Index of categories

Entries to also be published on truelite.it

Free as in Facebook

Yesterday we were in an airport. We tried to connect to the airport "free" wifi. It had a captive portal that asked for a lot of personal information before one could maybe get on the internet, and we gave up. Bologna Airport, no matter what they do to pretend that they like you, it's always clear that they don't.

I looked at the captive portal screen and I said: «ah yes, "free" wifi. Free as in Facebook».

We figured that we had an expression that will want to be reused.

Posted Mon Mar 9 10:58:49 2015 Tags:

Setting up Akonadi

Now that I have a CalDAV server that syncs with my phone I would like to use it from my desktop.

It looks like akonadi is able to sync with CalDAV servers, so I'm giving it a try.

First thing first is to give a meaning to the arbitrary name of this thing. Wikipedia says it is the oracle goddess of justice in Ghana. That still does not hint at all at personal information servers, but seems quite nice. Ok. I gave up with software having purpose-related names ages ago.

# apt-get install akonadi-server akonadi-backend-postgresql

Akonadi wants a SQL database as a backend. By default it uses MySQL, but I had enough of MySQL ages ago.

I tried SQLite but the performance with it is terrible. Terrible as in, it takes 2 minutes between adding a calendar entry and having it show up in the calendar. I'm fascinated by how Akonadi manages to use SQLite so badly, but since I currently just want to get a job done, next in line is PostgreSQL:

# su - postgres
$ createuser enrico
$ psql postgres
postgres=# alter user enrico createdb;

Then as enrico:

$ createdb akonadi-enrico
$ cat <<EOT > ~/.config/akonadi/akonadiserverrc
[%General]
Driver=QPSQL

[QPSQL]
Name=akonadi-enrico
StartServer=false
Host=
Options=
ServerPath=
InitDbPath=

I can now use kontact to connect Akonadi to my CalDAV server and it works nicely, both with calendar and with addressbook entries.

KDE has at least two clients for Akonadi: Kontact, which is a kitchen sink application similar to Evolution, and KOrganizer, which is just the calendar and scheduling component of Kontact.

Both work decently, and KOrganizer has a pretty decent startup time. I now have a usable desktop PIM application that is synced with my phone. W00T!

Next step is to port my swift little calendar display tool to use Akonadi as a back-end.

Posted Tue Feb 17 15:34:55 2015 Tags:

seat-inspect

Four months ago I wrote this somewhere:

Seeing a DD saying "this new dbus stuff scares me" would make most debian users scared. Seeing a DD who has an idea of what is going on, and who can explain it, would be an interesting and exciting experience.

So, let's be exemplary, competent and patient. Or at least, competent. Some may like or not like the changes, but do we all understand what is going on? Will we all be able to support our friends and customers running jessie?

I confess that although I understand the need for it, I don't feel competent enough to support systemd-based machines right now.

So, are we maybe in need of help, cheat sheets, arsenals of one-liners, diagnostic tools?

Maybe a round of posts on -planet like "one debian package a day" but with new features that jessie will have, and how to understand them and take advantage of them?

That was four months ago. In the meantime, I did some work, and it got better for me.

Yesterday, however, I've seen an experienced Linux person frustrated because the shutdown function of the desktop was doing nothing whatsoever. Today I found John Goerzen's post on planet.

I felt like some more diagnostic tools were needed, so I spent the day making seat-inspect.

seat-inspect tries to make the status of the login/seat system visible, to help with understanding and troubleshooting.

The intent of running the code is to have an overview of the system status, both to see what the new facilities are about, and to figure out if there is something out of place.

The intent of reading the code is to have an idea of how to use these facilities: the code has been written to be straightforward and is annotated with relevant bits from the logind API documentation.

seat-inspect is not a finished tool, but a starting point. I put it on github hoping that people will fork it and add their own extra sanity checks and warnings, so that it can grow into a standard thing to run if a system acts weird.

As it is now, it should be able to issue warnings if some bits are missing for network-manager or shutdown functions to work correctly. I haven't really tested that, though, because I don't have a system at hand where they are currently not working fine.

Another nice thing of it is that when running seat-inspect -v you get a dump of what logind/consolekit think about your system. I found it an interesting way to explore the new functionalities that we recently grew. The same can be done, and in more details, with loginctl calls, but I lacked a summary.

After writing this I feel a bit more competent, probably enough to sit at somebody's computer and poke into loginctl bits. I highly recommend the experience.

Posted Tue Feb 10 18:06:43 2015 Tags:

Playing with python, terminfo and command output

I am experimenting with showing progress on the terminal for a subcommand that is being run, showing what is happening without scrolling away the output of the main program, and I came out with this little toy. It shows the last X lines of a subcommand output, then gets rid of everything after the command has ended.

Usability-wise, it feels like a tease to me: it looks like I'm being shown all sorts of information then they are taken away from me before I managed to make sense of them. However, I find it cute enough to share:

#!/usr/bin/env python3
#coding: utf-8
# Copyright 2015 Enrico Zini <enrico@enricozini.org>.  Licensed under the terms
# of the GNU General Public License, version 2 or any later version.

import argparse
import fcntl
import select
import curses
import contextlib
import subprocess
import os
import sys
import collections
import shlex
import shutil
import logging

def stream_output(proc):
    """
    Take a subprocess.Popen object and generate its output, line by line,
    annotated with "stdout" or "stderr". At process termination it generates
    one last element: ("result", return_code) with the return code of the
    process.
    """
    fds = [proc.stdout, proc.stderr]
    bufs = [b"", b""]
    types = ["stdout", "stderr"]
    # Set both pipes as non-blocking
    for fd in fds:
        fcntl.fcntl(fd, fcntl.F_SETFL, os.O_NONBLOCK)
    # Multiplex stdout and stderr with different prefixes
    while len(fds) > 0:
        s = select.select(fds, (), ())
        for fd in s[0]:
            idx = fds.index(fd)
            buf = fd.read()
            if len(buf) == 0:
                fds.pop(idx)
                if len(bufs[idx]) != 0:
                    yield types[idx], bufs.pop(idx)
                types.pop(idx)
            else:
                bufs[idx] += buf
                lines = bufs[idx].split(b"\n")
                bufs[idx] = lines.pop()
                for l in lines:
                    yield types[idx], l
    res = proc.wait()
    yield "result", res

@contextlib.contextmanager
def miniscreen(has_fancyterm, name, maxlines=3, silent=False):
    """
    Show the output of a process scrolling in a portion of the screen.

    has_fancyterm: true if the terminal supports fancy features; if false, just
    write lines to standard output

    name: name of the process being run, to use as a header

    maxlines: maximum height of the miniscreen

    silent: do nothing whatsoever, used to disable this without needing to
            change the code structure

    Usage:
        with miniscreen(True, "my process", 5) as print_line:
            for i in range(10):
                print_line(("stdout", "stderr")[i % 2], "Line #{}".format(i))
    """
    if not silent and has_fancyterm:
        # Discover all the terminal control sequences that we need
        output_normal = str(curses.tigetstr("sgr0"), "ascii")
        output_up = str(curses.tigetstr("cuu1"), "ascii")
        output_clreol = str(curses.tigetstr("el"), "ascii")
        cols, lines = shutil.get_terminal_size()
        output_width = cols

        fg_color = (curses.tigetstr("setaf") or
                    curses.tigetstr("setf") or "")
        sys.stdout.write(str(curses.tparm(fg_color, 6), "ascii"))

        output_lines = collections.deque(maxlen=maxlines)

        def print_lines():
            """
            Print the lines in our buffer, then move back to the beginning
            """
            sys.stdout.write("{} progress:".format(name))
            sys.stdout.write(output_clreol)
            for msg in output_lines:
                sys.stdout.write("\n")
                sys.stdout.write(msg)
                sys.stdout.write(output_clreol)
            sys.stdout.write(output_up * len(output_lines))
            sys.stdout.write("\r")

        try:
            print_lines()

            def _progress_line(type, line):
                """
                Print a new line to the miniscreen
                """
                # Add the new line to our output buffer
                msg = "{} {}".format("." if type == "stdout" else "!", line)
                if len(msg) > output_width - 4:
                    msg = msg[:output_width - 4] + "..."
                output_lines.append(msg)
                # Update the miniscreen
                print_lines()

            yield _progress_line

            # Clear the miniscreen by filling our ring buffer with empty lines
            # then printing them out
            for i in range(maxlines):
                output_lines.append("")
            print_lines()
        finally:
            sys.stdout.write(output_normal)
    elif not silent:
        def _progress_line(type, line):
            print("{}: {}".format(type, line))
        yield _progress_line
    else:
        def _progress_line(type, line):
            pass
        yield _progress_line

def run_command_fancy(name, cmd, env=None, logfd=None, fancy=True, debug=False):
    quoted_cmd = " ".join(shlex.quote(x) for x in cmd)
    log.info("%s running command %s", name, quoted_cmd)
    if logfd: print("runcmd:", quoted_cmd, file=logfd)

    # Run the script itself on an empty environment, so that what was
    # documented is exactly what was run
    proc = subprocess.Popen(cmd, stdout=subprocess.PIPE, stderr=subprocess.PIPE, env=env)

    with miniscreen(fancy, name, silent=debug) as progress:
        stderr = []
        for type, val in stream_output(proc):
            if type == "stdout":
                val = val.decode("utf-8")
                if logfd: print("stdout:", val, file=logfd)
                log.debug("%s stdout: %s", name, val)
                progress(type, val)
            elif type == "stderr":
                val = val.decode("utf-8")
                if logfd: print("stderr:", val, file=logfd)
                stderr.append(val)
                log.debug("%s stderr: %s", name, val)
                progress(type, val)
            elif type == "result":
                if logfd: print("retval:", val, file=logfd)
                log.debug("%s retval: %d", name, val)
                retval = val

    if retval != 0:
        lastlines = min(len(stderr), 5)
        log.error("%s exited with code %s", name, retval)
        log.error("Last %d lines of standard error:", lastlines)
        for line in stderr[-lastlines:]:
            log.error("%s: %s", name, line)

    return retval


parser = argparse.ArgumentParser(description="run a command showing only a portion of its output")
parser.add_argument("--logfile", action="store", help="specify a file where the full execution log will be written")
parser.add_argument("--debug", action="store_true", help="debugging output on the terminal")
parser.add_argument("--verbose", action="store_true", help="verbose output on the terminal")
parser.add_argument("command", nargs="*", help="command to run")
args = parser.parse_args()

if args.debug:
    loglevel = logging.DEBUG
elif args.verbose:
    loglevel = logging.INFO
else:
    loglevel = logging.WARN
logging.basicConfig(level=loglevel, stream=sys.stderr)
log = logging.getLogger()

fancy = False
if not args.debug and sys.stdout.isatty():
    curses.setupterm()
    if curses.tigetnum("colors") > 0:
        fancy = True

if args.logfile:
    logfd = open("output.log", "wt")
else:
    logfd = None

retval = run_command_fancy("miniscreen example", args.command, logfd=logfd)

sys.exit(retval)
Posted Wed Jan 21 11:13:31 2015 Tags:

Upgrade Cyanogenmod with an encrypted phone

Cyanogenmod found an update, it downloaded it, then it rebooted to install it and nothing happened. It turns out that the update procedure cannot work if the zip file to install is in encrypted media, so a workaround is to move the zip into unencrypted external storage.

As far as I know, my Nexus 4 has no unencrypted external storage.

This is how I managed to upgrade it, I write it here so I can find it next time:

  1. enable USB debugging
  2. adb pull /cmupdater/cm-11-20141115-SNAPSHOT-M12-mako.zip
  3. adb reboot recovery
  4. choose "install zip from sideload"
  5. adb sideload cm-11-20141115-SNAPSHOT-M12-mako.zip
Posted Fri Dec 19 10:21:29 2014 Tags:

Radicale and DAVDroid

radicale and DAVdroid appeal to me. Let's try to make the whole thing work.

A self-signed SSL certificate

Generating the certificate:

    openssl req -nodes -x509 -newkey rsa:2048 -keyout cal-key.pem -out cal-cert.pem -days 3650
    [...]
    Country Name (2 letter code) [AU]:IT
    State or Province Name (full name) [Some-State]:Bologna
    Locality Name (eg, city) []:
    Organization Name (eg, company) [Internet Widgits Pty Ltd]:enricozini.org
    Organizational Unit Name (eg, section) []:
    Common Name (e.g. server FQDN or YOUR name) []:cal.enricozini.org
    Email Address []:postmaster@enricozini.org

Installing it on my phone:

    openssl x509 -in cal-cert.pem -outform DER -out cal-cert.crt
    adb push cal-cert.crt /mnt/sdcard/
    enrico --follow-instructions http://davdroid.bitfire.at/faq/entry/importing-a-certificate

Installing radicale in my VPS

An updated radicale package, with this patch to make it work with DAVDroid:

    apt-get source radicale
    # I reviewed 063f7de7a2c7c50de5fe3f8382358f9a1124fbb6
    git clone https://github.com/Kozea/Radicale.git
    Move the python code from git to the Debian source
    dch -v 0.10~enrico  "Pulled in the not yet released 0.10 work from upstream"
    debuild -us -uc -rfakeroot

Install the package:

    # dpkg -i python-radicale_0.10~enrico0-1_all.deb
    # dpkg -i radicale_0.10~enrico0-1_all.deb

Create a system user to run it:

    # adduser --system --disabled-password radicale

Configure it for mod_wsgi with auth done by Apache:

    # For brevity, this is my config file with comments removed

    [storage]
    # Storage backend
    # Value: filesystem | multifilesystem | database | custom
    type = filesystem

    # Folder for storing local collections, created if not present
    filesystem_folder = /var/lib/radicale/collections

    [logging]
    config = /etc/radicale/logging

Create the wsgi file to run it:

    # mkdir /srv/radicale
    # cat <<EOT > /srv/radicale/radicale.wsgi
    import radicale
    radicale.log.start()
    application = radicale.Application()
    EOT
    # chown radicale.radicale /srv/radicale/radicale.wsgi
    # chmod 0755 /srv/radicale/radicale.wsgi

Make radicale commit to git

    # apt-get install python-dulwich
    # cd /var/lib/radicale/collections
    # git init
    # chown radicale.radicale -R /var/lib/radicale/collections/.git

Apache configuration

Add a new site to apache:

    $ cat /etc/apache2/sites-available/cal.conf
    # For brevity, this is my config file with comments removed
    <IfModule mod_ssl.c>
    <VirtualHost *:443>
            ServerName cal.enricozini.org
            ServerAdmin enrico@enricozini.org

            Alias /robots.txt /srv/radicale/robots.txt
            Alias /favicon.ico /srv/radicale/favicon.ico

            WSGIDaemonProcess radicale user=radicale group=radicale threads=1 umask=0027 display-name=%{GROUP}
            WSGIProcessGroup radicale
            WSGIScriptAlias / /srv/radicale/radicale.wsgi

            <Directory /srv/radicale>
                    # WSGIProcessGroup radicale
                    # WSGIApplicationGroup radicale
                    # WSGIPassAuthorization On
                    AllowOverride None
                    Require all granted
            </Directory>

            <Location />
                    AuthType basic
                    AuthName "Enrico's Calendar"
                    AuthBasicProvider file
                    AuthUserFile /usr/local/etc/radicale/htpasswd
                    Require user enrico
            </Location>

            ErrorLog{APACHE_LOG_DIR}/cal-enricozini-org-error.log
            LogLevel warn

            CustomLog{APACHE_LOG_DIR}/cal-enricozini-org-access.log combined

            SSLEngine on
            SSLCertificateFile    /etc/ssl/certs/cal.pem
            SSLCertificateKeyFile /etc/ssl/private/cal.key
    </VirtualHost>
    </IfModule>

Then enable it:

    # a2ensite cal.conf
    # service apache2 reload

Create collections

DAVdroid seems to want to see existing collections on the server, so we create them:

    $ apt-get install cadaver
    $ cat <<EOT > /tmp/empty.ics
    BEGIN:VCALENDAR
    VERSION:2.0
    END:VCALENDAR
    EOT
    $ cat <<EOT > /tmp/empty.vcf
    BEGIN:VCARD
    VERSION:2.1
    END:VCARD
    EOT
    $ cadaver https://cal.enricozini.org
    WARNING: Untrusted server certificate presented for `cal.enricozini.org':
    [...]
    Do you wish to accept the certificate? (y/n) y
    Authentication required for Enrico's Calendar on server `cal.enricozini.org':
    Username: enrico
    Password: ****
    dav:/> cd enrico/contacts.vcf/
    dav:/> put /tmp/empty.vcf
    dav:/> cd ../calendar.ics/
    dav:/> put /tmp/empty.ics
    dav:/enrico/calendar.ics/> ^D
    Connection to `cal.enricozini.org' closed.

DAVdroid configuration

  1. Add a new DAVdroid sync account
  2. Use server/username configuration
  3. For server, use https:////
  4. Add username and password

It should work.

Related links

Posted Tue Dec 9 16:35:50 2014 Tags:

Alternate rescue boot entry with systemd

Since systemd version 215, adding systemd.debug-shell to the kernel command line activates the debug shell on tty9 alongside the normal boot. I like the idea of that, and I'd like to have it in my standard 'rescue' entry in my grub menu.

Unfortunately, by default update-grub does not allow to customize the rescue menu entry options. I have just filed #766530 hoping for that to change.

After testing the patch I proposed for /etc/grub.d/10_linux, I now have this in my /etc/default/grub, with some satisfaction:

GRUB_CMDLINE_LINUX_RECOVERY="systemd.log_target=kmsg systemd.log_level=debug systemd.debug-shell"

Further information:

Thanks to sjoerd and uau on #debian-systemd for their help.

Posted Thu Oct 23 22:06:30 2014 Tags:

Spelling a chilometri zero

Lo spelling internazionale è troppo globalizzato, e volete recuperare un attimo la dimensione del posto dove siete nati e cresciuti?

Da oggi c'è questo script che fa per voi: gli dite dove abitate, e lui vi crea lo spelling a chilometri zero.

$ git clone git@gitorious.org:trespolo/osmspell.git
$ cd osmspell
$ ./osmspell "San Giorgio di Piano"
1: San Giorgio di Piano, BO, EMR, Italia
2: San Giorgio di Piano, Via Codronchi, San Giorgio di Piano, BO, EMR, Italia
3: San Giorgio Di Piano, Via Libertà, San Giorgio di Piano, BO, EMR, Italia
Choose one: 1
Center: 44.6465332, 11.3790398
A Argelato, Altedo
B Bentivoglio, Bologna, Boschi
C Cinquanta, Castagnolo Minore, Castel Maggiore, Cento
D Dosso
E Eremo di Tizzano
F Funo di Argelato, Finale Emilia, Ferrara, Fiesso
G Gherghenzano, Galliera, Gesso
I Il Cucco, Irnerio, Idice
L Località Fortuna, Lovoleto, Lippo
M Malacappa, Massumatico, Minerbio, Marano
N Navile
O Osteriola, Ozzano dell'Emilia, Oca
P Piombino, Padulle, Poggio Renatico, Piave
Q Quarto Inferiore, Quattrina
R Rubizzano, Renazzo, Riale
S San Giorgio di Piano, Saletto
T Torre Verde, Tintoria, Tombe
U Uccellino
V Venezzano Mascarino, Vigarano Mainarda, Veduro
X XII Morelli
Z Zenerigolo, Zola Predosa

I dati vengono da OSM, e lo script è un ottimo esempio di come usarne la API di geolocazione (veloci) e la API di query geografica (lenta).

Posted Sat Jan 4 00:38:16 2014 Tags:
Posted Tue Apr 7 19:20:12 2015

Python-related posts.

Custom function decorators with TurboGears 2

I am exposing some library functions using a TurboGears2 controller (see web-api-with-turbogears2). It turns out that some functions return a dict, some a list, some a string, and TurboGears 2 only allows JSON serialisation for dicts.

A simple work-around for this is to wrap the function result into a dict, something like this:

@expose("json")
@validate(validator_dispatcher, error_handler=api_validation_error)
def list_colours(self, filter=None, productID=None, maxResults=100, **kw):
    # Call API
    res = self.engine.list_colours(filter, productID, maxResults)

    # Return result
    return dict(r=res)

It would be nice, however, to have an @webapi() decorator that automatically wraps the function result with the dict:

def webapi(func):
    def dict_wrap(*args, **kw):
        return dict(r=func(*args, **kw))
    return dict_wrap

# ...in the controller...

    @expose("json")
    @validate(validator_dispatcher, error_handler=api_validation_error)
    @webapi
    def list_colours(self, filter=None, productID=None, maxResults=100, **kw):
        # Call API
        res = self.engine.list_colours(filter, productID, maxResults)

        # Return result
        return res

This works, as long as @webapi appears last in the list of decorators. This is because if it appears last it will be the first to wrap the function, and so it will not interfere with the tg.decorators machinery.

Would it be possible to create a decorator that can be put anywhere among the decorator list? Yes, it is possible but tricky, and it gives me the feeling that it may break in any future version of TurboGears:

class webapi(object):
    def __call__(self, func):
        def dict_wrap(*args, **kw):
            return dict(r=func(*args, **kw))
        # Migrate the decoration attribute to our new function
        if hasattr(func, 'decoration'):
            dict_wrap.decoration = func.decoration
            dict_wrap.decoration.controller = dict_wrap
            delattr(func, 'decoration')
        return dict_wrap

# ...in the controller...

    @expose("json")
    @validate(validator_dispatcher, error_handler=api_validation_error)
    @webapi
    def list_colours(self, filter=None, productID=None, maxResults=100, **kw):
        # Call API
        res = self.engine.list_colours(filter, productID, maxResults)

        # Return result
        return res

As a convenience, TurboGears 2 offers, in the decorators module, a way to build decorator "hooks":

class before_validate(_hook_decorator):
    '''A list of callables to be run before validation is performed'''
    hook_name = 'before_validate'

class before_call(_hook_decorator):
    '''A list of callables to be run before the controller method is called'''
    hook_name = 'before_call'

class before_render(_hook_decorator):
    '''A list of callables to be run before the template is rendered'''
    hook_name = 'before_render'

class after_render(_hook_decorator):
    '''A list of callables to be run after the template is rendered.

    Will be run before it is returned returned up the WSGI stack'''

    hook_name = 'after_render'

The way these are invoked can be found in the _perform_call function in tg/controllers.py.

To show an example use of those hooks, let's add a some polygen wisdom to every data structure we return:

class wisdom(decorators.before_render):
    def __init__(self, grammar):
        super(wisdom, self).__init__(self.add_wisdom)
        self.grammar = grammar
    def add_wisdom(self, remainder, params, output):
        from subprocess import Popen, PIPE
        output["wisdom"] = Popen(["polyrun", self.grammar], stdout=PIPE).communicate()[0]

# ...in the controller...

    @wisdom("genius")
    @expose("json")
    @validate(validator_dispatcher, error_handler=api_validation_error)
    def list_colours(self, filter=None, productID=None, maxResults=100, **kw):
        # Call API
        res = self.engine.list_colours(filter, productID, maxResults)
    
        # Return result
        return res

These hooks cannot however be used for what I need, that is, to wrap the result inside a dict. The reason is because they are called in this way:

        controller.decoration.run_hooks(
                'before_render', remainder, params, output)

and not in this way:

        output = controller.decoration.run_hooks(
                'before_render', remainder, params, output)

So it is possible to modify the output (if it is a mutable structure) but not to exchange it with something else.

Can we do even better? Sure we can. We can assimilate @expose and @validate inside @webapi to avoid repeating those same many decorator lines over and over again:

class webapi(object):
    def __init__(self, error_handler = None):
        self.error_handler = error_handler

    def __call__(self, func):
        def dict_wrap(*args, **kw):
            return dict(r=func(*args, **kw))
        res = expose("json")(dict_wrap)
        res = validate(validator_dispatcher, error_handler=self.error_handler)(res)
        return res

# ...in the controller...

    @expose("json")
    def api_validation_error(self, **kw):
        pylons.response.status = "400 Error"
        return dict(e="validation error on input fields", form_errors=pylons.c.form_errors)

    @webapi(error_handler=api_validation_error)
    def list_colours(self, filter=None, productID=None, maxResults=100, **kw):
        # Call API
        res = self.engine.list_colours(filter, productID, maxResults)

        # Return result
        return res

This got rid of @expose and @validate, and provides almost all the default values that I need. Unfortunately I could not find out how to access api_validation_error from the decorator so that I can pass it to the validator, therefore I remain with the inconvenience of having to explicitly pass it every time.

Posted Wed Nov 4 17:52:38 2009 Tags:

Building a web-based API with Turbogears2

I am using TurboGears2 to export a python API over the web. Every API method is wrapper by a controller method that validates the parameters and returns the results encoded in JSON.

The basic idea is this:

@expose("json")
def list_colours(self, filter=None, productID=None, maxResults=100, **kw):
    # Call API
    res = self.engine.list_colours(filter, productID, maxResults)

    # Return result
    return res

To validate the parameters we can use forms, it's their job after all:

class ListColoursForm(TableForm):
    fields = [
            # One field per parameter
            twf.TextField("filter", help_text="Please enter the string to use as a filter"),
            twf.TextField("productID", help_text="Please enter the product ID"),
            twf.TextField("maxResults", validator=twfv.Int(min=0), default=200, size=5, help_text="Please enter the maximum number of results"),
    ]
list_colours_form=ListColoursForm()

#...

    @expose("json")
    @validate(list_colours_form, error_handler=list_colours_validation_error)
    def list_colours(self, filter=None, productID=None, maxResults=100, **kw):
        # Parameter validation is done by the form
    
        # Call API
        res = self.engine.list_colours(filter, productID, maxResults)
    
        # Return result
        return res

All straightforward so far. However, this means that we need two exposed methods for every API call: one for the API call and one error handler. For every API call, we have to type the name several times, which is error prone and risks to get things mixed up.

We can however have a single error handler for all methonds:

def get_method():
    '''
    The method name is the first url component after the controller name that
    does not start with 'test'
    '''
    found_controller = False
    for name in pylons.c.url.split("/"):
        if not found_controller and name == "controllername":
            found_controller = True
            continue
        if name.startswith("test"):
            continue
        if found_controller:
            return name
    return None

class ValidatorDispatcher:
    '''
    Validate using the right form according to the value of the "method" field
    '''
    def validate(self, args, state):
        method = args.get("method", None)
    # Extract the method from the URL if it is missing
        if method is None:
            method = get_method()
            args["method"] = method
        return forms[method].validate(args, state)

validator_dispatcher = ValidatorDispatcher()

This validator will try to find the method name, either as a form field or by parsing the URL. It will then use the method name to find the form to use for validation, and pass control to the validate method of that form.

We then need to add an extra "method" field to our forms, and arrange the forms inside a dictionary:

class ListColoursForm(TableForm):
    fields = [
            # One hidden field to have a place for the method name
            twf.HiddenField("method")
            # One field per parameter
            twf.TextField("filter", help_text="Please enter the string to use as a filter"),
    #...

forms["list_colours"] = ListColoursForm()

And now our methods become much nicer to write:

    @expose("json")
    def api_validation_error(self, **kw):
        pylons.response.status = "400 Error"
        return dict(form_errors=pylons.c.form_errors)

    @expose("json")
    @validate(validator_dispatcher, error_handler=api_validation_error)
    def list_colours(self, filter=None, productID=None, maxResults=100, **kw):
        # Parameter validation is done by the form
    
        # Call API
        res = self.engine.list_colours(filter, productID, maxResults)
    
        # Return result
        return res

api_validation_error is interesting: it returns a proper HTTP error status, and a JSON body with the details of the error, taken straight from the form validators. It took me a while to find out that the form errors are in pylons.c.form_errors (and for reference, the form values are in pylons.c.form_values). pylons.response is a WebOb Response that we can play with.

So now our client side is able to call the API methods, and get a proper error if it calls them wrong.

But now that we have the forms ready, it doesn't take much to display them in web pages as well:

def _describe(self, method):
    "Return a dict describing an API method"
    ldesc = getattr(self.engine, method).__doc__.strip()
    sdesc = ldesc.split("\n")[0]
    return dict(name=method, sdesc = sdesc, ldesc = ldesc)

@expose("myappserver.templates.myappapi")
def index(self):
    '''
    Show an index of exported API methods
    '''
    methods = dict()
    for m in forms.keys():
        methods[m] = self._describe(m)
    return dict(methods=methods)

@expose('myappserver.templates.testform')
def testform(self, method, **kw):
    '''
    Show a form with the parameters of an API method
    '''
    kw["method"] = method
    return dict(method=method, action="/myapp/test/"+method, value=kw, info=self._describe(method), form=forms[method])

@expose(content_type="text/plain")
@validate(validator_dispatcher, error_handler=testform)
def test(self, method, **kw):
    '''
    Run an API method and show its prettyprinted result
    '''
    res = getattr(self, str(method))(**kw)
    return pprint.pformat(res)

In a few lines, we have all we need: an index of the API methods (including their documentation taken from the docstrings!), and for each method a form to invoke it and a page to see the results.

Make the forms children of AjaxForm, and you can even see the results together with the form.

Posted Thu Oct 15 15:45:39 2009 Tags:

Creating pipelines with subprocess

It is possible to create process pipelines using subprocess.Popen, by just using stdout=subprocess.PIPE and stdin=otherproc.stdout.

Almost.

In a pipeline created in this way, the stdout of all processes except the last is opened twice: once in the script that has run the subprocess and another time in the standard input of the next process in the pipeline.

This is a problem because if a process closes its stdin, the previous process in the pipeline does not get SIGPIPE when trying to write to its stdout, because that pipe is still open on the caller process. If this happens, a wait on that process will hang forever: the child process waits for the parent to read its stdout, the parent process waits for the child process to exit.

The trick is to close the stdout of each process in the pipeline except the last just after creating them:

#!/usr/bin/python
# coding=utf-8

import subprocess

def pipe(*args):
    '''
    Takes as parameters several dicts, each with the same
    parameters passed to popen.

    Runs the various processes in a pipeline, connecting
    the stdout of every process except the last with the
    stdin of the next process.
    '''
    if len(args) < 2:
        raise ValueError, "pipe needs at least 2 processes"
    # Set stdout=PIPE in every subprocess except the last
    for i in args[:-1]:
        i["stdout"] = subprocess.PIPE

    # Runs all subprocesses connecting stdins and stdouts to create the
    # pipeline. Closes stdouts to avoid deadlocks.
    popens = [subprocess.Popen(**args[0])]
    for i in range(1,len(args)):
        args[i]["stdin"] = popens[i-1].stdout
        popens.append(subprocess.Popen(**args[i]))
        popens[i-1].stdout.close()

    # Returns the array of subprocesses just created
    return popens

At this point, it's nice to write a function that waits for the whole pipeline to terminate and returns an array of result codes:

def pipe_wait(popens):
    '''
    Given an array of Popen objects returned by the
    pipe method, wait for all processes to terminate
    and return the array with their return values.
    '''
    results = [0] * len(popens)
    while popens:
        last = popens.pop(-1)
        results[len(popens)] = last.wait()
    return results

And, look and behold, we can now easily run a pipeline and get the return codes of every single process in it:

process1 = dict(args='sleep 1; grep line2 testfile', shell=True)
process2 = dict(args='awk \'{print $3}\'', shell=True)
process3 = dict(args='true', shell=True)
popens = pipe(process1, process2, process3)
result = pipe_wait(popens)
print result

Update: Colin Watson suggests an improvement to compensate for Python's nonstandard SIGPIPE handling.

Colin Watson has a similar library for C.

Posted Wed Jul 1 09:08:06 2009 Tags:

Linking to self in turbogears

I want to put in my master.kid some icons that allow to change the current language for the session.

First, all user-accessible methods need to handle a 'language' parameter:

@expose(template="myapp.templates.foobar")
def index(self, someparam, **kw):
    if 'language' in kw: turbogears.i18n.set_session_locale(kw['language'])

Then, we need a way to edit the current URL so that we can generate modified links to self that preserve the existing path_info and query parameters. In your main controller, add:

def linkself(**kw):
    params = {}
    params.update(cherrypy.request.params)
    params.update(kw)
    url = cherrypy.request.browser_url.split('?', 1)[0]
    return url + '?' + '&'.join(['='.join(x) for x in params.iteritems()])

def add_custom_stdvars(vars):
    return vars.update({"linkself": linkself})

turbogears.view.variable_providers.append(add_custom_stdvars)

(see the turbogears stdvars documentation and the cherrypy request documentation (cherrypy 2 documentation at the bottom of the page))

And finally, in master.kid:

<div id="footer">
  <div id="langselector">
    <span class="language">
      <a href="${tg.linkself(language='it_IT')}">
        <img src="${tg.url('/static/images/it.png')}"/>
      </a>
    </span>

    <span class="language">
      <a href="${tg.linkself(language='C')}">
        <img src="${tg.url('/static/images/en.png')}"/>
      </a>
    </span>
  </div><!-- langselector -->
</div><!-- footer -->
Posted Sat Jun 6 00:57:39 2009 Tags:

Turbogears quirks when testing controllers that use SingleSelectField

Suppose you have a User that can be a member of a Company. In SQLObject you model it somehow like this:

    class Company(SQLObject):
        name = UnicodeCol(length=16, alternateID=True, alternateMethodName="by_name")
        display_name = UnicodeCol(length=255)

    class User(InheritableSQLObject):
        company = ForeignKey("Company", notNull=False, cascade='null')

Then you want to make a form that allows to choose what is the company of a user:

def companies():
    return [ [ -1, 'None' ] ] + [ [c.id, c.display_name] for c in Company.select() ]

class NewUserFields(WidgetsList):
    """Fields for editing general settings"""
    user_name = TextField(label="User name")
    companyID = SingleSelectField(label="Company", options=companies)

Ok. Now you want to run tests:

  1. nosetests imports the controller to see if there's any initialisation code.
  2. The NewUserFields class is created.
  3. The SingleSelectField is created.
  4. The SingleSelectField constructor tries to guess the validator and peeks at the first option.
  5. This calls companies.
  6. companies accesses the database.
  7. The testing database has not yet been created because nosetests imported the module before giving the test code a chance to setup the test database.
  8. Bang.

The solution is to add an explicit validator to disable this guessing code that is a source of so many troubles:

class NewUserFields(WidgetsList):
    """Fields for editing general settings"""
    user_name = TextField(label="User name")
    companyID = SingleSelectField(label="Company", options=companies, validator=v.Int(not_empty=True))
Posted Sat Jun 6 00:57:39 2009 Tags:

Quirks when overriding SQLObject setters

Let's suppose you have a User that is, optionally, a member of a Company. In SQLObject you model it somehow like this:

    class Company(SQLObject):
        name = UnicodeCol(length=16, alternateID=True, alternateMethodName="by_name")
        display_name = UnicodeCol(length=255)

    class User(InheritableSQLObject):
        company = ForeignKey("Company", notNull=False, cascade='null')

Then you want to implement a user settings interface that uses a Select box to choose the company of the user.

For the Select widget to properly handle the validator for your data, you need to put a number in the first option. As my first option, I want to have the "None" entry, so I decided to use -1 to mean "None".

Now, to make it all blend nicely, I overrode the company setter to accept -1 and silently convert it to a None:

    class User(InheritableSQLObject):
        company = ForeignKey("Company", notNull=False, cascade='null')

        def _set_company(self, id):
            "Set the company id, using None if -1 is given"
            if id == -1: id = None
            self._SO_set_company(id)

In the controller, after parsing and validating all the various keyword arguments, I do something like this:

            user.set(**kw)

Now, the overridden method didn't get called.

After some investigation, and with the help of NandoFlorestan on IRC, we figured out the following things:

  1. That method needs to be rewritten as _set_companyID:

            def _set_companyID(self, id):
                "Set the company id, using None if -1 is given"
                if id == -1: id = None
                self._SO_set_companyID(id)
    
  2. Methods overridden in that way are alsop called by user.set(**kw), but not by the User(**kw) constructor, so using, for example, a similar override to transparently encrypt passwords would give you plaintext passwords for new users and encrypted passwords after they changed it.

Posted Sat Jun 6 00:57:39 2009 Tags:

Turbogears form quirk

I had a great idea:

@validate(model_form)
@error_handler()
@expose(template='kid:myproject.templates.new')
def new(self, id, tg_errors=None, **kw):
    """Create new records in model"""
    if tg_errors:
        # Ask until there is still something missing
        return dict(record = defaults, form = model_form)
    else:
        # We have everything: save it
        i = Item(**kw)
        flash("Item was successfully created.")
        raise redirect("../show/%d" % i.id)

It was perfect: one simple method, simple error handling, nice helpful messages all around. Except, check boxes and select fields would not get the default values while all other fields would.

After two hours searching and cursing and tracing things into widget code, I found this bit in InputWidget.adjust_value:

# there are some input fields that when nothing is checked/selected
# instead of sending a nice name="" are totally missing from
# input_values, this little workaround let's us manage them nicely
# without interfering with other types of fields, we need this to
# keep track of their empty status otherwise if the form is going to be
# redisplayed for some errors they end up to use their defaults values
# instead of being empty since FE doesn't validate a failing Schema.
# posterity note: this is also why we need if_missing=None in
# validators.Schema, see ticket #696.

So, what is happening here is that since check boxes and option fields don't have a nice behaviour when unselected, turbogears has to work around it. So in order to detect the difference between "I selected 'None'" and "I didn't select anything", it reasons that if the input has been validated, then the user has made some selections, so it defaults to "The user selected 'None'". If the input has not been validated, then we're showing the form for the first time, then a missing value means "Use the default provided".

Since I was doing the validation all the time, this meant that Checkboxes and Select fields would never use the default values.

Hence, if you use those fields then you necessarily need two different controller methods, one to present the form and one to save it:

@expose(template='kid:myproject.templates.new')
def new(self, id, **kw):
    """Create new records in model"""
    return dict(record = defaults(), form = model_form)

@validate(model_form)
@error_handler(new)
@expose()
def savenew(self, id, **kw):
    """Create new records in model"""
    i = Item(**kw)
    flash("Item was successfully created.")
    raise redirect("../show/%d"%i.id)

If someone else stumbles on the same problem, I hope they'll find this post and they won't have to spend another two awful hours tracking it down again.

Posted Sat Jun 6 00:57:39 2009 Tags:

Turbogears i18n quirks

Collecting strings from .kid files

tg-admin i18n collect won't collect strings from your .kid files: you need the toolbox web interface for that.

Indentation problems in .kid files

The toolbox web interface chokes on intentation errors on your .kid files.

To see the name of the .kid file that causes the error, look at the tg-admin toolbox output in the terminal for lines like Working on app/Foo/templates/bar.kid.

What happens is that the .kid files are converted to python files, and if there are indentation glitches they end up in the python files, and python will complain.

Once you see from the tg-admin toolbox standard error what is the .kid file with the problem, edit it and try to make sure that all closing tags are at the exact indentation level as their coresponding opening tags. Even a single space would matter.

Bad i18n bug in TurboKid versions earlier than 1.0.1

faide on #turbogears also says:

It is of outmost importance that you use TurboKid 1.0.1 because it is the first version that corrects a BIG bug regarding i18n filters ...

The version below had a bug where the filters kept being added at each page load in such a way that after a few hundreds of pages you could have page loading times as long as 5 minutes!

If one has a previous version of TurboKid, one (and only one) of these is needed:

So, in short, all i18n users should upgrade to TurboGears 1.0.2.2 or patch TurboKid using http://trac.turbogears.org/ticket/1301.

Posted Sat Jun 6 00:57:39 2009 Tags:

File downloads with TurboGears

In TurboGears, I had to implement a file download method, but the file required access controls so it was put in a directory not exported by Apache.

In #turbogears I've been pointed at: http://cherrypy.org/wiki/FileDownload and this is everything put together:

from cherrypy.lib.cptools import serveFile
# In cherrypy 3 it should be:
#from cherrypy.lib.static import serve_file

@expose()
def get(self, *args, **kw):
    """Access the file pointed by the given path"""
    pathname = check_auth_and_compute_pathname()
    return serveFile(pathname)

Then I needed to export some CSV:

@expose()
def getcsv(self, *args, **kw):
    """Get the data in CSV format"""
    rows = compute_data_rows()
    headers = compute_headers(rows)
    filename = compute_file_name()

    cherrypy.response.headers['Content-Type'] = "application/x-download"
    cherrypy.response.headers['Content-Disposition'] = 'attachment; filename="'+filename+'"'

    csvdata = StringIO.StringIO()
    writer = csv.writer(csvdata)
    writer.writerow(headers)
    writer.writerows(rows)

    return csvdata.getvalue()

In my case it's not an issue as I can only compute the headers after I computed all the data, but I still have to find out how to serve the CSV file while I'm generating it, instead of storing it all into a big string and returning the big string.

Posted Sat Jun 6 00:57:39 2009 Tags:

TurboGears RemoteForm tip

In case your RemoteForm misteriously behaves like a normal HTTP form, refreshing the page on submit, and the only hint that there's something wrong is this bit in the Iceweasel's error console:

Errore: uncaught exception: [Exception... "Component returned failure
code: 0x80070057 (NS_ERROR_ILLEGAL_VALUE) [nsIXMLHttpRequest.open]"
nsresult: "0x80070057 (NS_ERROR_ILLEGAL_VALUE)"  location: "JS frame ::
javascript: eval(__firebugTemp__); :: anonymous :: line 1"  data: no]

the problem can just be a missing action= attribute to the form.

I found out after:

  1. reading the TurboGears remoteform wiki: "For some reason, the RemoteForm is acting like a regular html form, serving up a new page instead of performing the replacements we're looking for. I'll update this page as soon as I figure out why this is happening."

  2. finding this page on Google and meditating for a while while staring at it. I don't speak German, but often enough I manage to solve problems after meditating over Google results in all sorts of languages unknown or unreadable to me. I will call this practice Webomancy.

Posted Sat Jun 6 00:57:39 2009 Tags:
Posted Sat Jun 6 00:57:39 2009

Pages about Turbogears.

Custom function decorators with TurboGears 2

I am exposing some library functions using a TurboGears2 controller (see web-api-with-turbogears2). It turns out that some functions return a dict, some a list, some a string, and TurboGears 2 only allows JSON serialisation for dicts.

A simple work-around for this is to wrap the function result into a dict, something like this:

@expose("json")
@validate(validator_dispatcher, error_handler=api_validation_error)
def list_colours(self, filter=None, productID=None, maxResults=100, **kw):
    # Call API
    res = self.engine.list_colours(filter, productID, maxResults)

    # Return result
    return dict(r=res)

It would be nice, however, to have an @webapi() decorator that automatically wraps the function result with the dict:

def webapi(func):
    def dict_wrap(*args, **kw):
        return dict(r=func(*args, **kw))
    return dict_wrap

# ...in the controller...

    @expose("json")
    @validate(validator_dispatcher, error_handler=api_validation_error)
    @webapi
    def list_colours(self, filter=None, productID=None, maxResults=100, **kw):
        # Call API
        res = self.engine.list_colours(filter, productID, maxResults)

        # Return result
        return res

This works, as long as @webapi appears last in the list of decorators. This is because if it appears last it will be the first to wrap the function, and so it will not interfere with the tg.decorators machinery.

Would it be possible to create a decorator that can be put anywhere among the decorator list? Yes, it is possible but tricky, and it gives me the feeling that it may break in any future version of TurboGears:

class webapi(object):
    def __call__(self, func):
        def dict_wrap(*args, **kw):
            return dict(r=func(*args, **kw))
        # Migrate the decoration attribute to our new function
        if hasattr(func, 'decoration'):
            dict_wrap.decoration = func.decoration
            dict_wrap.decoration.controller = dict_wrap
            delattr(func, 'decoration')
        return dict_wrap

# ...in the controller...

    @expose("json")
    @validate(validator_dispatcher, error_handler=api_validation_error)
    @webapi
    def list_colours(self, filter=None, productID=None, maxResults=100, **kw):
        # Call API
        res = self.engine.list_colours(filter, productID, maxResults)

        # Return result
        return res

As a convenience, TurboGears 2 offers, in the decorators module, a way to build decorator "hooks":

class before_validate(_hook_decorator):
    '''A list of callables to be run before validation is performed'''
    hook_name = 'before_validate'

class before_call(_hook_decorator):
    '''A list of callables to be run before the controller method is called'''
    hook_name = 'before_call'

class before_render(_hook_decorator):
    '''A list of callables to be run before the template is rendered'''
    hook_name = 'before_render'

class after_render(_hook_decorator):
    '''A list of callables to be run after the template is rendered.

    Will be run before it is returned returned up the WSGI stack'''

    hook_name = 'after_render'

The way these are invoked can be found in the _perform_call function in tg/controllers.py.

To show an example use of those hooks, let's add a some polygen wisdom to every data structure we return:

class wisdom(decorators.before_render):
    def __init__(self, grammar):
        super(wisdom, self).__init__(self.add_wisdom)
        self.grammar = grammar
    def add_wisdom(self, remainder, params, output):
        from subprocess import Popen, PIPE
        output["wisdom"] = Popen(["polyrun", self.grammar], stdout=PIPE).communicate()[0]

# ...in the controller...

    @wisdom("genius")
    @expose("json")
    @validate(validator_dispatcher, error_handler=api_validation_error)
    def list_colours(self, filter=None, productID=None, maxResults=100, **kw):
        # Call API
        res = self.engine.list_colours(filter, productID, maxResults)
    
        # Return result
        return res

These hooks cannot however be used for what I need, that is, to wrap the result inside a dict. The reason is because they are called in this way:

        controller.decoration.run_hooks(
                'before_render', remainder, params, output)

and not in this way:

        output = controller.decoration.run_hooks(
                'before_render', remainder, params, output)

So it is possible to modify the output (if it is a mutable structure) but not to exchange it with something else.

Can we do even better? Sure we can. We can assimilate @expose and @validate inside @webapi to avoid repeating those same many decorator lines over and over again:

class webapi(object):
    def __init__(self, error_handler = None):
        self.error_handler = error_handler

    def __call__(self, func):
        def dict_wrap(*args, **kw):
            return dict(r=func(*args, **kw))
        res = expose("json")(dict_wrap)
        res = validate(validator_dispatcher, error_handler=self.error_handler)(res)
        return res

# ...in the controller...

    @expose("json")
    def api_validation_error(self, **kw):
        pylons.response.status = "400 Error"
        return dict(e="validation error on input fields", form_errors=pylons.c.form_errors)

    @webapi(error_handler=api_validation_error)
    def list_colours(self, filter=None, productID=None, maxResults=100, **kw):
        # Call API
        res = self.engine.list_colours(filter, productID, maxResults)

        # Return result
        return res

This got rid of @expose and @validate, and provides almost all the default values that I need. Unfortunately I could not find out how to access api_validation_error from the decorator so that I can pass it to the validator, therefore I remain with the inconvenience of having to explicitly pass it every time.

Posted Wed Nov 4 17:52:38 2009 Tags:

Building a web-based API with Turbogears2

I am using TurboGears2 to export a python API over the web. Every API method is wrapper by a controller method that validates the parameters and returns the results encoded in JSON.

The basic idea is this:

@expose("json")
def list_colours(self, filter=None, productID=None, maxResults=100, **kw):
    # Call API
    res = self.engine.list_colours(filter, productID, maxResults)

    # Return result
    return res

To validate the parameters we can use forms, it's their job after all:

class ListColoursForm(TableForm):
    fields = [
            # One field per parameter
            twf.TextField("filter", help_text="Please enter the string to use as a filter"),
            twf.TextField("productID", help_text="Please enter the product ID"),
            twf.TextField("maxResults", validator=twfv.Int(min=0), default=200, size=5, help_text="Please enter the maximum number of results"),
    ]
list_colours_form=ListColoursForm()

#...

    @expose("json")
    @validate(list_colours_form, error_handler=list_colours_validation_error)
    def list_colours(self, filter=None, productID=None, maxResults=100, **kw):
        # Parameter validation is done by the form
    
        # Call API
        res = self.engine.list_colours(filter, productID, maxResults)
    
        # Return result
        return res

All straightforward so far. However, this means that we need two exposed methods for every API call: one for the API call and one error handler. For every API call, we have to type the name several times, which is error prone and risks to get things mixed up.

We can however have a single error handler for all methonds:

def get_method():
    '''
    The method name is the first url component after the controller name that
    does not start with 'test'
    '''
    found_controller = False
    for name in pylons.c.url.split("/"):
        if not found_controller and name == "controllername":
            found_controller = True
            continue
        if name.startswith("test"):
            continue
        if found_controller:
            return name
    return None

class ValidatorDispatcher:
    '''
    Validate using the right form according to the value of the "method" field
    '''
    def validate(self, args, state):
        method = args.get("method", None)
    # Extract the method from the URL if it is missing
        if method is None:
            method = get_method()
            args["method"] = method
        return forms[method].validate(args, state)

validator_dispatcher = ValidatorDispatcher()

This validator will try to find the method name, either as a form field or by parsing the URL. It will then use the method name to find the form to use for validation, and pass control to the validate method of that form.

We then need to add an extra "method" field to our forms, and arrange the forms inside a dictionary:

class ListColoursForm(TableForm):
    fields = [
            # One hidden field to have a place for the method name
            twf.HiddenField("method")
            # One field per parameter
            twf.TextField("filter", help_text="Please enter the string to use as a filter"),
    #...

forms["list_colours"] = ListColoursForm()

And now our methods become much nicer to write:

    @expose("json")
    def api_validation_error(self, **kw):
        pylons.response.status = "400 Error"
        return dict(form_errors=pylons.c.form_errors)

    @expose("json")
    @validate(validator_dispatcher, error_handler=api_validation_error)
    def list_colours(self, filter=None, productID=None, maxResults=100, **kw):
        # Parameter validation is done by the form
    
        # Call API
        res = self.engine.list_colours(filter, productID, maxResults)
    
        # Return result
        return res

api_validation_error is interesting: it returns a proper HTTP error status, and a JSON body with the details of the error, taken straight from the form validators. It took me a while to find out that the form errors are in pylons.c.form_errors (and for reference, the form values are in pylons.c.form_values). pylons.response is a WebOb Response that we can play with.

So now our client side is able to call the API methods, and get a proper error if it calls them wrong.

But now that we have the forms ready, it doesn't take much to display them in web pages as well:

def _describe(self, method):
    "Return a dict describing an API method"
    ldesc = getattr(self.engine, method).__doc__.strip()
    sdesc = ldesc.split("\n")[0]
    return dict(name=method, sdesc = sdesc, ldesc = ldesc)

@expose("myappserver.templates.myappapi")
def index(self):
    '''
    Show an index of exported API methods
    '''
    methods = dict()
    for m in forms.keys():
        methods[m] = self._describe(m)
    return dict(methods=methods)

@expose('myappserver.templates.testform')
def testform(self, method, **kw):
    '''
    Show a form with the parameters of an API method
    '''
    kw["method"] = method
    return dict(method=method, action="/myapp/test/"+method, value=kw, info=self._describe(method), form=forms[method])

@expose(content_type="text/plain")
@validate(validator_dispatcher, error_handler=testform)
def test(self, method, **kw):
    '''
    Run an API method and show its prettyprinted result
    '''
    res = getattr(self, str(method))(**kw)
    return pprint.pformat(res)

In a few lines, we have all we need: an index of the API methods (including their documentation taken from the docstrings!), and for each method a form to invoke it and a page to see the results.

Make the forms children of AjaxForm, and you can even see the results together with the form.

Posted Thu Oct 15 15:45:39 2009 Tags:

Linking to self in turbogears

I want to put in my master.kid some icons that allow to change the current language for the session.

First, all user-accessible methods need to handle a 'language' parameter:

@expose(template="myapp.templates.foobar")
def index(self, someparam, **kw):
    if 'language' in kw: turbogears.i18n.set_session_locale(kw['language'])

Then, we need a way to edit the current URL so that we can generate modified links to self that preserve the existing path_info and query parameters. In your main controller, add:

def linkself(**kw):
    params = {}
    params.update(cherrypy.request.params)
    params.update(kw)
    url = cherrypy.request.browser_url.split('?', 1)[0]
    return url + '?' + '&'.join(['='.join(x) for x in params.iteritems()])

def add_custom_stdvars(vars):
    return vars.update({"linkself": linkself})

turbogears.view.variable_providers.append(add_custom_stdvars)

(see the turbogears stdvars documentation and the cherrypy request documentation (cherrypy 2 documentation at the bottom of the page))

And finally, in master.kid:

<div id="footer">
  <div id="langselector">
    <span class="language">
      <a href="${tg.linkself(language='it_IT')}">
        <img src="${tg.url('/static/images/it.png')}"/>
      </a>
    </span>

    <span class="language">
      <a href="${tg.linkself(language='C')}">
        <img src="${tg.url('/static/images/en.png')}"/>
      </a>
    </span>
  </div><!-- langselector -->
</div><!-- footer -->
Posted Sat Jun 6 00:57:39 2009 Tags:

Turbogears quirks when testing controllers that use SingleSelectField

Suppose you have a User that can be a member of a Company. In SQLObject you model it somehow like this:

    class Company(SQLObject):
        name = UnicodeCol(length=16, alternateID=True, alternateMethodName="by_name")
        display_name = UnicodeCol(length=255)

    class User(InheritableSQLObject):
        company = ForeignKey("Company", notNull=False, cascade='null')

Then you want to make a form that allows to choose what is the company of a user:

def companies():
    return [ [ -1, 'None' ] ] + [ [c.id, c.display_name] for c in Company.select() ]

class NewUserFields(WidgetsList):
    """Fields for editing general settings"""
    user_name = TextField(label="User name")
    companyID = SingleSelectField(label="Company", options=companies)

Ok. Now you want to run tests:

  1. nosetests imports the controller to see if there's any initialisation code.
  2. The NewUserFields class is created.
  3. The SingleSelectField is created.
  4. The SingleSelectField constructor tries to guess the validator and peeks at the first option.
  5. This calls companies.
  6. companies accesses the database.
  7. The testing database has not yet been created because nosetests imported the module before giving the test code a chance to setup the test database.
  8. Bang.

The solution is to add an explicit validator to disable this guessing code that is a source of so many troubles:

class NewUserFields(WidgetsList):
    """Fields for editing general settings"""
    user_name = TextField(label="User name")
    companyID = SingleSelectField(label="Company", options=companies, validator=v.Int(not_empty=True))
Posted Sat Jun 6 00:57:39 2009 Tags:

Quirks when overriding SQLObject setters

Let's suppose you have a User that is, optionally, a member of a Company. In SQLObject you model it somehow like this:

    class Company(SQLObject):
        name = UnicodeCol(length=16, alternateID=True, alternateMethodName="by_name")
        display_name = UnicodeCol(length=255)

    class User(InheritableSQLObject):
        company = ForeignKey("Company", notNull=False, cascade='null')

Then you want to implement a user settings interface that uses a Select box to choose the company of the user.

For the Select widget to properly handle the validator for your data, you need to put a number in the first option. As my first option, I want to have the "None" entry, so I decided to use -1 to mean "None".

Now, to make it all blend nicely, I overrode the company setter to accept -1 and silently convert it to a None:

    class User(InheritableSQLObject):
        company = ForeignKey("Company", notNull=False, cascade='null')

        def _set_company(self, id):
            "Set the company id, using None if -1 is given"
            if id == -1: id = None
            self._SO_set_company(id)

In the controller, after parsing and validating all the various keyword arguments, I do something like this:

            user.set(**kw)

Now, the overridden method didn't get called.

After some investigation, and with the help of NandoFlorestan on IRC, we figured out the following things:

  1. That method needs to be rewritten as _set_companyID:

            def _set_companyID(self, id):
                "Set the company id, using None if -1 is given"
                if id == -1: id = None
                self._SO_set_companyID(id)
    
  2. Methods overridden in that way are alsop called by user.set(**kw), but not by the User(**kw) constructor, so using, for example, a similar override to transparently encrypt passwords would give you plaintext passwords for new users and encrypted passwords after they changed it.

Posted Sat Jun 6 00:57:39 2009 Tags:

Turbogears form quirk

I had a great idea:

@validate(model_form)
@error_handler()
@expose(template='kid:myproject.templates.new')
def new(self, id, tg_errors=None, **kw):
    """Create new records in model"""
    if tg_errors:
        # Ask until there is still something missing
        return dict(record = defaults, form = model_form)
    else:
        # We have everything: save it
        i = Item(**kw)
        flash("Item was successfully created.")
        raise redirect("../show/%d" % i.id)

It was perfect: one simple method, simple error handling, nice helpful messages all around. Except, check boxes and select fields would not get the default values while all other fields would.

After two hours searching and cursing and tracing things into widget code, I found this bit in InputWidget.adjust_value:

# there are some input fields that when nothing is checked/selected
# instead of sending a nice name="" are totally missing from
# input_values, this little workaround let's us manage them nicely
# without interfering with other types of fields, we need this to
# keep track of their empty status otherwise if the form is going to be
# redisplayed for some errors they end up to use their defaults values
# instead of being empty since FE doesn't validate a failing Schema.
# posterity note: this is also why we need if_missing=None in
# validators.Schema, see ticket #696.

So, what is happening here is that since check boxes and option fields don't have a nice behaviour when unselected, turbogears has to work around it. So in order to detect the difference between "I selected 'None'" and "I didn't select anything", it reasons that if the input has been validated, then the user has made some selections, so it defaults to "The user selected 'None'". If the input has not been validated, then we're showing the form for the first time, then a missing value means "Use the default provided".

Since I was doing the validation all the time, this meant that Checkboxes and Select fields would never use the default values.

Hence, if you use those fields then you necessarily need two different controller methods, one to present the form and one to save it:

@expose(template='kid:myproject.templates.new')
def new(self, id, **kw):
    """Create new records in model"""
    return dict(record = defaults(), form = model_form)

@validate(model_form)
@error_handler(new)
@expose()
def savenew(self, id, **kw):
    """Create new records in model"""
    i = Item(**kw)
    flash("Item was successfully created.")
    raise redirect("../show/%d"%i.id)

If someone else stumbles on the same problem, I hope they'll find this post and they won't have to spend another two awful hours tracking it down again.

Posted Sat Jun 6 00:57:39 2009 Tags:

Turbogears i18n quirks

Collecting strings from .kid files

tg-admin i18n collect won't collect strings from your .kid files: you need the toolbox web interface for that.

Indentation problems in .kid files

The toolbox web interface chokes on intentation errors on your .kid files.

To see the name of the .kid file that causes the error, look at the tg-admin toolbox output in the terminal for lines like Working on app/Foo/templates/bar.kid.

What happens is that the .kid files are converted to python files, and if there are indentation glitches they end up in the python files, and python will complain.

Once you see from the tg-admin toolbox standard error what is the .kid file with the problem, edit it and try to make sure that all closing tags are at the exact indentation level as their coresponding opening tags. Even a single space would matter.

Bad i18n bug in TurboKid versions earlier than 1.0.1

faide on #turbogears also says:

It is of outmost importance that you use TurboKid 1.0.1 because it is the first version that corrects a BIG bug regarding i18n filters ...

The version below had a bug where the filters kept being added at each page load in such a way that after a few hundreds of pages you could have page loading times as long as 5 minutes!

If one has a previous version of TurboKid, one (and only one) of these is needed:

So, in short, all i18n users should upgrade to TurboGears 1.0.2.2 or patch TurboKid using http://trac.turbogears.org/ticket/1301.

Posted Sat Jun 6 00:57:39 2009 Tags:

File downloads with TurboGears

In TurboGears, I had to implement a file download method, but the file required access controls so it was put in a directory not exported by Apache.

In #turbogears I've been pointed at: http://cherrypy.org/wiki/FileDownload and this is everything put together:

from cherrypy.lib.cptools import serveFile
# In cherrypy 3 it should be:
#from cherrypy.lib.static import serve_file

@expose()
def get(self, *args, **kw):
    """Access the file pointed by the given path"""
    pathname = check_auth_and_compute_pathname()
    return serveFile(pathname)

Then I needed to export some CSV:

@expose()
def getcsv(self, *args, **kw):
    """Get the data in CSV format"""
    rows = compute_data_rows()
    headers = compute_headers(rows)
    filename = compute_file_name()

    cherrypy.response.headers['Content-Type'] = "application/x-download"
    cherrypy.response.headers['Content-Disposition'] = 'attachment; filename="'+filename+'"'

    csvdata = StringIO.StringIO()
    writer = csv.writer(csvdata)
    writer.writerow(headers)
    writer.writerows(rows)

    return csvdata.getvalue()

In my case it's not an issue as I can only compute the headers after I computed all the data, but I still have to find out how to serve the CSV file while I'm generating it, instead of storing it all into a big string and returning the big string.

Posted Sat Jun 6 00:57:39 2009 Tags:

TurboGears RemoteForm tip

In case your RemoteForm misteriously behaves like a normal HTTP form, refreshing the page on submit, and the only hint that there's something wrong is this bit in the Iceweasel's error console:

Errore: uncaught exception: [Exception... "Component returned failure
code: 0x80070057 (NS_ERROR_ILLEGAL_VALUE) [nsIXMLHttpRequest.open]"
nsresult: "0x80070057 (NS_ERROR_ILLEGAL_VALUE)"  location: "JS frame ::
javascript: eval(__firebugTemp__); :: anonymous :: line 1"  data: no]

the problem can just be a missing action= attribute to the form.

I found out after:

  1. reading the TurboGears remoteform wiki: "For some reason, the RemoteForm is acting like a regular html form, serving up a new page instead of performing the replacements we're looking for. I'll update this page as soon as I figure out why this is happening."

  2. finding this page on Google and meditating for a while while staring at it. I don't speak German, but often enough I manage to solve problems after meditating over Google results in all sorts of languages unknown or unreadable to me. I will call this practice Webomancy.

Posted Sat Jun 6 00:57:39 2009 Tags:

Passing values to turbogears widgets at display time

In turbogears, I often need to pass data to widgets at display time. Sometimes it works automatically, but sometimes, in cases like passing option lists to CheckBoxLists or number of repetitions in a RepeatingFieldSet, it doesn't.

All the examples use precomputed lists or pass simple code functions. In most of my cases, I want them computed by the controller every time.

Passing a function hasn't worked, as I did not find any obvious way to have the function know about the controller.

So I need to pass things the display() method of the widgets, but I could not work out how to pass the option list and default list for a CheckBoxList that is part of a WidgetsList in a TableForm.

On IRC came the answer, thanks to Xentac:

you should be able to...
    tableform.display(options=dict(checkboxname=[optionlist]))

And yes, it works. I can pass the default value as one of the normal form values:

    tableform.display(values=dict(checkboxname=[values]), options=dict(checkboxname=[optionlist]))
Posted Sat Jun 6 00:57:39 2009 Tags:
Posted Sat Jun 6 00:57:39 2009

Rants, kept to a bare minimum and strictly with a useful component.

Another day in the life of a poor developer

try:
    # After Python 3.3
    from collections.abc import Iterable
except ImportError:
    # This has changed in Python 3.3 (why, oh why?), reinforcing the idea that
    # the best Python version ever is still 2.7, simply because upstream has
    # promised that they won't touch it (and break it) for at least 5 more
    # years.
    from collections import Iterable

import shlex
if hasattr(shlex, "quote"):
    # New in version 3.3.
    shell_quote = shlex.quote
else:
    # Available since python 1.6 but deprecated since version 2.7: Prior to Python
    # 2.7, this function was not publicly documented. It is finally exposed
    # publicly in Python 3.3 as the quote function in the shlex module.
    #
    # Except everyone was using it, because it was the only way provided by the
    # python standard library to make a string safe for shell use
    #
    # See http://stackoverflow.com/questions/35817/how-to-escape-os-system-calls-in-python
    import pipes
    shell_quote = pipes.quote

import shutil
if hasattr(shutil, "which"):
    # New in version 3.3.
    shell_which = shutil.which
else:
    # Available since python 1.6:
    # http://stackoverflow.com/questions/377017/test-if-executable-exists-in-python
    from distutils.spawn import find_executable
    shell_which = find_executable
Posted Fri Feb 27 12:02:33 2015 Tags:

Akonadi client example

After many failed attemps I have managed to build a C++ akonadi client. It has felt like one of the most frustrating programming experiences of my whole life, so I'm sharing the results hoping to spare others from all the suffering.

First thing first, akonadi client libraries are not in libakonadi-dev but in kdepimlibs5-dev, even if kdepimlibs5-dev does not show in apt-cache search akonadi.

Then, kdepimlibs is built with Qt4. If your application uses Qt5 (mine was) you need to port it back to Qt4 if you want to talk to Akonadi.

Then, kdepimlibs does not seem to support qmake and does not ship pkg-config .pc files, and if you want to use kdepimlibs your build system needs to be cmake. I ported by code from qmake to cmake, and now qtcreator wants me to run cmake by hand every time I change the CMakeLists.txt file, and it stopped allowing to add, rename or delete sources.

Finally, most of the code / build system snippets found on the internet seem flawed in a way or another, because the build toolchain of Qt/KDE applications has undergone several redesignins during time, and the network is littered with examples from different eras. The way to obtain template code to start a Qt/KDE project is to use kapptemplate. I have found no getting started tutorial on the internet that said "do not just copy the snippets from here, run kapptemplate instead so you get them up to date".

kapptemplate supports building an "Akonadi Resource" and an "Akonadi Serializer", but it does not support generating template code for an akonadi client. That left me with the feeling that I was dealing with some software that wants to be developed but does not want to be used.

Anyway, now an example of how to interrogate Akonadi exists as is on the internet. I hope that all the tears of blood that I cried this morning have not been cried in vain.

Posted Mon Feb 23 15:44:01 2015 Tags:

The wonders of missing documentation

Update: I have managed to build an example Akonadi client application.

I'm new here, I want to make a simple C++ GUI app that pops up a QCalendarWidget which my local Akonadi has appointments.

I open qtcreator, create a new app, hack away for a while, then of course I get undefined references for all Akonadi symbols, since I didn't tell the build system that I'm building with akonadi. Ok.

How do I tell the build system that I'm building with akonadi? After 20 minutes of frantic looking around the internet, I still have no idea.

There is a package called libakonadi-dev which does not seem to have anything to do with this. That page mentions everything about making applications with Akonadi except how to build them.

There is a package called kdepimlibs5-dev which looks promising: it has no .a files but it does have haders and cmake files. However, qtcreator is only integrated with qmake, and I would really like the handholding of an IDE at this stage.

I put something together naively doing just what looked right, and I managed to get an application that segfaults before main() is even called:

/*
 * Copyright © 2015 Enrico Zini <enrico@enricozini.org>
 *
 * This work is free. You can redistribute it and/or modify it under the
 * terms of the Do What The Fuck You Want To Public License, Version 2,
 * as published by Sam Hocevar. See the COPYING file for more details.
 */
#include <QDebug>

int main(int argc, char *argv[])
{
    qDebug() << "BEGIN";
    return 0;
}
QT       += core gui widgets
CONFIG += c++11

TARGET = wtf
TEMPLATE = app

LIBS += -lkdecore -lakonadi-kde

SOURCES += wtf.cpp

I didn't achieve what I wanted, but I feel like I achieved something magical and beautiful after all.

I shall now perform some haruspicy on those oscure cmake files to see if I can figure something out. But seriously, people?

Posted Mon Feb 23 11:36:18 2015 Tags:

Mozilla marketplace facepalm

This made me sad.

My view, which didn't seem to be considered in that discussion, is that people concerned about software freedom and security are likely to stay the hell away from such an app market and its feedback forms.

Also, that thread made me so sad about the state of that developer community that I seriously do not feel like investing energy into going through the hoops of getting an account in their bugtracker to point this out.

Sigh.

Posted Fri Jan 23 15:13:16 2015 Tags:

SSL "protection"

In my experience with my VPS, setting up pretty much any service exposed to the internet, even a simple thing to put a calendar in my phone requires an SSL certificate, which costs money, which needs to be given to some corporation or another.

When the only way to get protection from a threat is to give money to some big fish, I feel like I'm being forced to pay protection money.

I look forward to this.

Posted Thu Dec 11 15:35:25 2014 Tags:

Laptop, I demand that you suspend!

Dear Lazyweb,

Sometimes some application prevents suspend on my laptop. I want to disable that feature: how?

I understand that there may exist some people who like that feature. I, on the other hand, consider a scenario like this inconceivable:

  1. I'm on a plane working with my laptop, the captain announces preparations for landing, so I quickly hit the suspend button (or close the lid) on my laptop and stow it away.
  2. One connecting flight later, I pick up my backpack, I feel it unusually hot and realise that my laptop has been on all along, and is now dead from either running out of battery or thermal protection.
  3. I think things that, if spoken aloud in front of a pentacle, might invoke major lovecraftian horrors.

I do not want this scenario to ever be possible. I want my suspend button to suspend the laptop no matter what. If a process does not agree, I'm fine with suspending it anyway, or killing it.

If I want my laptop to suspend, I generally have a good enough real-world reason for it, and I cannot conceive that a software could ever be allowed to override my command.

How do I change this? I don't know if I should look into systemd, upowerd, pm-utils, the kernel, the display manager or something else entirely. I worry that I cannot even figure where to start looking for a solution.

This happened to me multiple times already, and I consider it ridiculous. I know that it can cause me data loss. I know that it can cause me serious trouble in case I was relying on having some battery or state left at my arrival. I know that depending on what is in my backpack, this could also be physically dangerous.

So, what knob do I tweak for this? How do I make suspend reliable?

Update

Systemd has an inhibitor system, and systemd-inhibit --list only lists 'delay' blocks in my system. It is an interesting feature that seems to be implemented in the right way, and it could mean that I finally can get my screen to be locked before the system is suspended.

It is possible to configure the inhibitor system in /etc/systemd/logind.conf, including ways to ignore inhibitors, and a maximum time after which inhibitors are ignored if not yet released.

Try as I might to run everything that I was running on the plane that time, I could not manage to see anything take an inhibitor block that could have prevented my suspend. I now suspect that what happened to me was a glitch caused by something else (hardware? kernel? cosmic rays!) during that specific suspend.

When I had this issue in the past it looks like the infrastructure at the time was far more primitive that what we have now with systemd, so I guess that when writing my blog post I had simply correlated my old experiences with a one-off suspend glitch.

If I want to investigate or tune further, to test the situation with a runaway block, I can use commands like systemd-inhibit --mode=block sleep 3600.

I'm quite happy to see that we're moving to a standard and sane system for this. In the meantime, I have learnt that pm-utils has now become superfluous and can be deinstalled, and so can acpi-support and acpi-support-base.

Thanks vbernat, mbiebl, and ah, on #debian-devel for all the help.

Posted Thu Sep 11 14:32:40 2014 Tags:

Wheezy for industrial software development

I'm helping with setting up a wheezy-based toolchain for industrial automation.

The basic requirements are: live-build, C++11, Qt 5.3, and a frozen internal wheezy mirror.

debmirror

A good part of a day's work was lost because of #749734 and possibly #628779. Mirror rebuild is still ongoing, and fingers crossed.

This is Italy, and you can't simply download 21Gb of debs just to see how it goes.

C++11

Stable toolchains for C++11 now exist and have gained fast adoption. It makes sense, since given what is in C++11 it is unthinkable to start a new C++ project with the old standard nowadays.

C++11 is supported by g++ 4.8+ or clang 3.3+. None of them is available on wheezy or wheezy-backports.

Backports exist of g++ 4.8 only for Ubuntu 12.04, but they are uninstallable on wheezy due at least to a different libc6. I tried rebuilding g++4.8 on wheezy but quickly gave up.

clang 3.3 has a build dependency on g++ 4.8. LOL.

However, LLVM provides an APT repository with their most recent compiler, and it works, too. C++11 problem solved!

Qt 5.3

Qt 5.3 is needed because of the range of platforms it can target. There is no wheezy backport that I can find.

I cannot simply get it from Qt's Download page and install it, since we need it packaged, to build live ISOs with it.

I'm attempting to backport the packages from experimental to wheezy.

Here are its build dependencies:

libxcb-1.10 (needed by qt5)

Building this is reasonably straightforward.

libxkbcommon 0.4.0 (needed by qt5)

The version from jessie builds fine on wheezy, provided you remove --fail-missing from the dh_install invocation.

libicu 52.1 (needed by harfbuzz)

The jessie packages build on wheezy, provided that mentions of clang are deleted from source/configure.ac, since it fails to build with clang 3.5 (the one currently available for wheezy on llvm.org).

libharfbuzz-dev

Backporting this is a bloodbath: the Debian packages from jessie depend on a forest of gobject hipsterisms of doom, all unavailable on wheezy. I gave up.

qt 5.3

qtbase-opensource-src-5.3.0+dfsg can be made to build with an embedded version of harfbuzz, with just this change:

diff -Naur a/debian/control a/debian/control
--- a/debian/control    2014-05-20 18:48:27.000000000 +0200
+++ b/debian/control    2014-05-29 17:45:31.037215786 +0200
@@ -28,7 +28,6 @@
                libgstreamer-plugins-base0.10-dev,
                libgstreamer0.10-dev,
                libgtk2.0-dev,
-               libharfbuzz-dev,
                libicu-dev,
                libjpeg-dev,
                libmysqlclient-dev,
diff -Naur a/debian/rules b/debian/rules
--- a/debian/rules  2014-05-18 01:56:37.000000000 +0200
+++ b/debian/rules  2014-05-29 17:45:25.738634371 +0200
@@ -108,7 +108,6 @@
                -plugin-sql-tds \
                -system-sqlite \
                -platform $(platform_arg) \
-               -system-harfbuzz \
                -system-zlib \
                -system-libpng \
                -system-libjpeg \

(thanks Lisandro Damián Nicanor Pérez Meyer for helping me there!)

There are probably going to be further steps in the Qt5 toolchain.

Actually, let's try prebuilt binaries

The next day with a fresh mind we realised that it is preferable to reduce our tampering with the original wheezy to a minimum. Our current plan is to use wheezy's original Qt and Qt-using packages, and use Qt's prebuilt binaries in /opt for all our custom software.

We run Qt's installer, tarred the result, and wrapped it in a Debian package like this:

$ cat debian/rules
#!/usr/bin/make -f

QT_VERSION = 5.3

%:
    dh $@

override_dh_auto_build:
    dh_auto_build
    sed -re 's/@QT_VERSION@/$(QT_VERSION)/g' debian-rules.inc.in > debian-rules.inc

override_dh_auto_install:
    dh_auto_install
    # Download and untar the prebuild Qt5 binaries
    install -d -o root -g root -m 0755 debian/our-qt5-sdk/opt/Qt
    curl http://localserver/Qt$(QT_VERSION).tar.xz | xz -d | tar -C debian/our-qt5-sdk/opt -xf -
    # Move the runtime part to our-qt5
    install -d -o root -g root -m 0755 debian/our-qt5/opt/Qt
    mv debian/our-qt5-sdk/opt/Qt/$(QT_VERSION) debian/our-qt5/opt/Qt/
    # Makes dpkg-shlibdeps work on packages built with Qt from /opt
    # Hack. Don't try this at home. Don't ever do this unless you
    # know what you are doing. This voids your warranty. If you
    # know what you are doing, you won't do this.
    find debian/our-qt5/opt/Qt/$(QT_VERSION)/gcc_64/lib -maxdepth 1 -type f -name "lib*.so*" \
        | sed -re 's,^.+/(lib[^.]+)\.so.+$$,\1 5 our-qt5 (>= $(QT_VERSION)),' > debian/our-qt5.shlibs


$ cat debian-rules.inc.in
export PATH := /opt/Qt/@QT_VERSION@/gcc_64/bin:$(PATH)
export QMAKESPEC=/opt/Qt/@QT_VERSION@/gcc_64/mkspecs/linux-clang/

To build one of our packages using Qt5.3 and clang, we just add this to its debian/rules:

include /usr/share/our-qt5/debian-rules.inc

Wrap up

We got the dependencies sorted. Hopefully the mirror will rebuild itself tonight and tomorrow we can resume working on our custom live system.

Posted Thu May 29 18:05:17 2014 Tags:

Shops

Christmas songs should only ever be played on Christmas day.

In church.

At midnight.

Unless I happen to be there.

Posted Mon Dec 2 14:07:58 2013 Tags:

Airports

Photo of a commercial in London City airport saying 'In the lap of luxury - Want to reach a captive audience with dwell time? Why advertise anywhere else? - London City Airport Media Sales'

In the airport, we are not travellers. We are a captive audience with dwell time.

In other words, suckers stuck in a room where the only pastime provided is spending money and staring at advertisements selling advertisement space in rooms full of suckers like them.

Posted Fri Nov 22 18:58:00 2013 Tags:

A vision wanted

Today Richard Stallman mailed all Italian LUGs asking that tomorrow's LinuxDay be called "GNU/Linux Day" instead.

I wonder how that is ever going to help a community so balkanised, that the only way Italian LUGs manage to do something together, is to say "let's not care what we all do, let's just do it on the same day and call it a national event". Of course a few LUGs still make a point of not doing anything on that day, because you know, Judean People's Front. Cawk.

Today a friend asked me if I could help her support people in installing Citrix Whatsit to set up a video conference to network meetings that will take place in a month in different cities. Those meetings are something I look forwad to. It wasn't much of a problem to say "no, I can't do that"; it was a problem to be unable to come up with some viable, Free alternatives.

I sometimes have to use Skype to talk with friends who also are Debian Developers, because I still haven't managed to make VoIP work unless I go through a commercial proxy.

There was the happy news that our regional administration is switching from MS Word to OpenOffice. It soon became a flamewar, because some people loudly complained that they should have used LibreOffice instead.

At DebConf, after spending an hour getting frustrated with the default formatting of bullet points in WhateverOffice Impress, I did my Debian Contributors talk using a text file in vim. And it was a success! Thanks Francois Marier for maintaining cowsay.

I can't sync contact lists and appointments between my N900, which runs a Debian derivative, and my laptop, because I don't want to have a Google account, and nothing else would work out of the box.

I don't even know how to keep a shared calendar with other DDs, without using a 3rd party cloud service that I don't want to trust with my life's personal schedule.

I need to do a code review of every vim plugin I need to use, because you can only get them by cloning GitHub repositories over plain http, and they don't even provide signed tags. That code is run with my own privileges every time I start a text editor, which is, like, all the time. I'm frightened at the idea of how many people blissfully don't think about what that means. Vim users. Programmers. Cool chaps.

Yet the most important thing in Debian today seems to be yet another crusade between upstart and systemd.

But we haven't had a lengthy discussion on why, although the excellent OpenStreetMap exists and many of us contribute to it, it seems to still be more immediate to hit Google Maps to get a route computed. How can we change that?

We haven't had a lengthy discussion on what can we offer to allow anyone to set up some social platform that won't get swamped with spam the first day and cracked open the second; that would allow people to share some photos with their friends only, and some with the rest of the world; that won't required a full-time paid person to maintain. That won't be obsolete and require a migration to a new platform in a year. That isn't Facebook or Google Plus.

I stopped taking photos because it's too much work to show them to people. Other people use Instagram. Whatever the hipster trend is for photo sharing today, October 25, 2013, I'm pretty sure it's not a Free platform.

But we can do something. We technology leaders. We are those who drive technological change!

For example, today I invested two hours of hard effort trying to figure out why libbuffy's test suite fails on kfreebsd. All while wondering why I was doing that, since I know all buffy's users personally, and none of them uses kfreebsd.

And I will take a day off work to study the library symbols file specification, so that next time I'll know right away if the new version of a C++ compiler decides that a template-generated symbol isn't worth adding to a library anymore.

What is this effort really about? It sometimes feel like micromanaging to me.

It's good to have excellent quality standards. But not without a vision.

Not until "reliable network printing with all PDF viewers and print servers we ship" is among our release goals.

Not until we commit to making sure that "sharing files between Debian users" will work out of the box, without the need of going through a 3rd party website, or email.

I'm not interested in spending energy discussing init systems. I'm interested in spending energy sharing stories of what cool stuff we can do in Debian today, out of the box. And what cool stuff we'll be able to do tomorrow.

Let's spend time on IRC, on mailing lists, and at the next Debian events, talking about why we are really into this. Talking about a vision!

Note: Please don't spend time telling me how to fix the problems I mentioned above. I'm not interested in help fixing some problems for me today. I'm interested in asking help fixing problems for everybody, right in the next stable release.

Remember, remember, the 5th of November, 2014.

Posted Fri Oct 25 22:38:11 2013 Tags:
Posted Sat Jun 6 00:57:39 2009
eng

Pages in English.

Love thy neighbor as thyself

‘Love thy neighbor as thyself’, words which astoundingly occur already in the Old Testament.

One can love one’s neighbor less than one loves oneself; one is then the egoist, the racketeer, the capitalist, the bourgeois. and although one may accumulate money and power one does not of necessity have a joyful heart, and the best and most attractive pleasures of the soul are blocked.

Or one can love one’s neighbor more than oneself—then one is a poor devil, full of inferiority complexes, with a longing to love everything and still full of hate and torment towards oneself, living in a hell of which one lays the fire every day anew.

But the equilibrium of love, the capacity to love without being indebted to anyone, is the love of oneself which is not taken away from any other, this love of one’s neighbor which does no harm to the self.

(From Herman Hesse, "My Belief")

I always have a hard time finding this quote on the Internet. Let's fix that.

Posted Wed May 20 11:35:15 2015 Tags:

Work around Google evil .ics feeds

I've happily been using 2015/akonadi-install for my calendars, and yesterday I added an .ics feed export from Google, as a URL file source. It is a link in the form: https://www.google.com/calendar/ical/person%40gmail.com/private-12341234123412341234123412341234/basic.ics

After doing that, I noticed that the fan in my laptop was on more often than usual, and I noticed that akonadi-server and postgres were running very often, and doing quite a lot of processing.

The evil

I investigated and realised that Google seems to be doing everything they can to make their ical feeds hard to sync against efficiently. This is the list of what I have observed Gmail doing to an unchanged ical feed:

  • Date: headers in HTTP replies are always now
  • If-Modified-Since: is not supported
  • DTSTAMP of each element is always now
  • VTIMEZONE entries appear in random order
  • ORGANIZER CN entries randomly change between full name and plus.google.com user ID
  • ATTENDEE entries randomly change between having a CN or not having it
  • TRIGGER entries change spontaneously
  • CREATED entries change spontaneously

This causes akonadi to download and reprocess the entire ical feed at every single poll, and I can't blame akonadi for doing it. In fact, Google is saying that there is a feed with several years worth of daily appointments that all keep being changed all the time.

The work-around

As a work-around, I have configured the akonadi source to point at a local file on disk, and I have written a script to update the file only if the .ics feed has actually changed.

Have a look at the script: I consider it far from trivial, since it needs to do a partial parsing of the .ics feed to throw away all the nondeterminism that Google pollutes it with.

The setup

The script needs to be run periodically, and I used it as an opportunity to try systemd user timers:

    $ cat ~/.config/systemd/user/update-ical-feeds.timer
    [Unit]
    Description=Updates ical feeds every hour
    # Only run when on AC power
    ConditionACPower=yes

    [Timer]
    # Run every hour
    OnActiveSec=1h
    # Run a minute after boot
    OnBootSec=1m
    Unit=update-ical-feeds.service

    $ cat ~/.config/systemd/user/update-ical-feeds.service
    [Unit]
    Description=Update ICal feeds

    [Service]
    # Use oneshot to prevent two updates being run in case the previous one
    # runs for more time than the timer interval
    Type=oneshot
    ExecStart=/home/enrico/tmp/calendars/update

    $ systemctl --user start update-ical-feeds.timer
    $ systemctl --user list-timers
    NEXT                         LEFT       LAST                         PASSED UNIT                    ACTIVATES
    Wed 2015-03-25 22:19:54 CET  59min left Wed 2015-03-25 21:19:54 CET  2s ago update-ical-feeds.timer update-ical-feeds.service

    1 timers listed.
    Pass --all to see loaded but inactive timers, too.

To reload the configuration after editing: systemctl --user daemon-reload.

Further investigation

I wonder if ConditionACPower needs to be in the .timer or in the .service, since there is a [Unit] section is in both. Update: I have been told it can be in the .timer.

I also wonder if there is a way to have the timer trigger only when online. There is a network-online.target and I do not know if it is applicable. I also do not know how to ask systemd if all the preconditions are currently met for a .service/.timer to run.

Finally, I especially wonder if it is worth hoping that Google will ever make their .ics feeds play nicely with calendar clients.

Posted Wed Mar 25 21:50:21 2015 Tags:

Screen-dependent window geometry

I have an external monitor for my laptop in my work desk at home, and when I work I keep a few windows like IRC on my laptop screen, and everything else on the external monitor. Then maybe I transfer on the sofa to watch a movie or in the kitchen to cook, and I unplug from the external monitor to bring the laptop with me. Then maybe I go back to the external monitor to resume working.

The result of this (with openbox) is that when I disconnect the external monitor all the windows on my external monitor get moved to the right edge of the laptop monitor, and when I reconnect the external monitor I need to rearrange them all again.

I would like to implement something that does the following:

  1. it keeps a dictionary mapping screen geometry to window geometries
  2. every time a window geometry and virtual desktop number changes, it gets recorded in the hash for the current screen geometry
  3. every time the screen geometry changes, for each window, if there was a saved window geometry + wirtual desktop number for it for the new screen geometry, it gets restored.

Questions:

  1. Is anything like this already implemented? Where?
  2. If not, what would be a convenient way to implement it myself, ideally in a wmctrl-like way that does not depend on a specific WM?

Note: I am not interested in switching to a different WM unless it is openbox with this feature implemented in it.

Posted Mon Mar 16 21:29:36 2015 Tags:

Reuse passwords in /etc/crypttab

Today's scenario was a laptop with an SSD and a spinning disk, and the goal was to deploy a Debian system on it so that as many things as possible are encrypted.

My preferred option for it is to setup one big LUKS partition in each disk, and put a LVM2 Physical Volume inside each partition. At boot, the two LUKS partition are opened, their contents are assembled into a Volume Group, and I can have everything I want inside.

This has advantages:

  • if any of the disks breaks, the other can still be unlocked, and it should still be possible to access the LVs inside it
  • once boot has happened, any layout of LVs can be used with no further worries about encryption
  • I can use pvmove to move partitions at will between SSD and spinning disks, which means I can at anytime renegotiate the tradeoffs between speed and disk space.

However, by default this causes cryptsetup to ask for the password once for each LUKS partition, even if the passwords are the same.

Searching for ways to mitigate this gave me unsatisfactory results, like:

  • decrypt the first disk, and use a file inside it as the keyfile to decrypt the second one. But in this case if the first disk breaks, I also lose the data in the second disk.
  • reuse the LUKS session key for the first disk in the second one. Same problem as before.
  • put a detached LUKS header in /boot and use it for both disks, then make regular backups of /boot. It is an interesting option that I have not tried.

The solution that I found was something that did not show up in any of my search results, so I'm documenting it here:

    # <target name> <source device>   <key file>   <options>
    ssd             /dev/sda2         main         luks,initramfs,discard,keyscript=decrypt_keyctl
    spin            /dev/sdb1         main         luks,initramfs,keyscript=decrypt_keyctl

This caches each password for 60 seconds, so that it can be reused to unlock other devices that use it. The documentation can be found at the beginning of /lib/cryptsetup/scripts/decrypt_keyctl, beware of the leopard™.

main is an arbitrary tag used to specify which devices use the same password.

This is also useful to work easily with multiple LUKS-on-LV setups:

    # <target name> <source device>          <key file>  <options>
    home            /dev/mapper/myvg-chome   main        luks,discard,keyscript=decrypt_keyctl
    backup          /dev/mapper/myvg-cbackup main        luks,discard,keyscript=decrypt_keyctl
    swap            /dev/mapper/myvg-cswap   main        swap,discard,keyscript=decrypt_keyctl
Posted Thu Mar 12 22:45:57 2015 Tags:

Free as in Facebook

Yesterday we were in an airport. We tried to connect to the airport "free" wifi. It had a captive portal that asked for a lot of personal information before one could maybe get on the internet, and we gave up. Bologna Airport, no matter what they do to pretend that they like you, it's always clear that they don't.

I looked at the captive portal screen and I said: «ah yes, "free" wifi. Free as in Facebook».

We figured that we had an expression that will want to be reused.

Posted Mon Mar 9 10:58:49 2015 Tags:

Another day in the life of a poor developer

try:
    # After Python 3.3
    from collections.abc import Iterable
except ImportError:
    # This has changed in Python 3.3 (why, oh why?), reinforcing the idea that
    # the best Python version ever is still 2.7, simply because upstream has
    # promised that they won't touch it (and break it) for at least 5 more
    # years.
    from collections import Iterable

import shlex
if hasattr(shlex, "quote"):
    # New in version 3.3.
    shell_quote = shlex.quote
else:
    # Available since python 1.6 but deprecated since version 2.7: Prior to Python
    # 2.7, this function was not publicly documented. It is finally exposed
    # publicly in Python 3.3 as the quote function in the shlex module.
    #
    # Except everyone was using it, because it was the only way provided by the
    # python standard library to make a string safe for shell use
    #
    # See http://stackoverflow.com/questions/35817/how-to-escape-os-system-calls-in-python
    import pipes
    shell_quote = pipes.quote

import shutil
if hasattr(shutil, "which"):
    # New in version 3.3.
    shell_which = shutil.which
else:
    # Available since python 1.6:
    # http://stackoverflow.com/questions/377017/test-if-executable-exists-in-python
    from distutils.spawn import find_executable
    shell_which = find_executable
Posted Fri Feb 27 12:02:33 2015 Tags:

Akonadi client example

After many failed attemps I have managed to build a C++ akonadi client. It has felt like one of the most frustrating programming experiences of my whole life, so I'm sharing the results hoping to spare others from all the suffering.

First thing first, akonadi client libraries are not in libakonadi-dev but in kdepimlibs5-dev, even if kdepimlibs5-dev does not show in apt-cache search akonadi.

Then, kdepimlibs is built with Qt4. If your application uses Qt5 (mine was) you need to port it back to Qt4 if you want to talk to Akonadi.

Then, kdepimlibs does not seem to support qmake and does not ship pkg-config .pc files, and if you want to use kdepimlibs your build system needs to be cmake. I ported by code from qmake to cmake, and now qtcreator wants me to run cmake by hand every time I change the CMakeLists.txt file, and it stopped allowing to add, rename or delete sources.

Finally, most of the code / build system snippets found on the internet seem flawed in a way or another, because the build toolchain of Qt/KDE applications has undergone several redesignins during time, and the network is littered with examples from different eras. The way to obtain template code to start a Qt/KDE project is to use kapptemplate. I have found no getting started tutorial on the internet that said "do not just copy the snippets from here, run kapptemplate instead so you get them up to date".

kapptemplate supports building an "Akonadi Resource" and an "Akonadi Serializer", but it does not support generating template code for an akonadi client. That left me with the feeling that I was dealing with some software that wants to be developed but does not want to be used.

Anyway, now an example of how to interrogate Akonadi exists as is on the internet. I hope that all the tears of blood that I cried this morning have not been cried in vain.

Posted Mon Feb 23 15:44:01 2015 Tags:

The wonders of missing documentation

Update: I have managed to build an example Akonadi client application.

I'm new here, I want to make a simple C++ GUI app that pops up a QCalendarWidget which my local Akonadi has appointments.

I open qtcreator, create a new app, hack away for a while, then of course I get undefined references for all Akonadi symbols, since I didn't tell the build system that I'm building with akonadi. Ok.

How do I tell the build system that I'm building with akonadi? After 20 minutes of frantic looking around the internet, I still have no idea.

There is a package called libakonadi-dev which does not seem to have anything to do with this. That page mentions everything about making applications with Akonadi except how to build them.

There is a package called kdepimlibs5-dev which looks promising: it has no .a files but it does have haders and cmake files. However, qtcreator is only integrated with qmake, and I would really like the handholding of an IDE at this stage.

I put something together naively doing just what looked right, and I managed to get an application that segfaults before main() is even called:

/*
 * Copyright © 2015 Enrico Zini <enrico@enricozini.org>
 *
 * This work is free. You can redistribute it and/or modify it under the
 * terms of the Do What The Fuck You Want To Public License, Version 2,
 * as published by Sam Hocevar. See the COPYING file for more details.
 */
#include <QDebug>

int main(int argc, char *argv[])
{
    qDebug() << "BEGIN";
    return 0;
}
QT       += core gui widgets
CONFIG += c++11

TARGET = wtf
TEMPLATE = app

LIBS += -lkdecore -lakonadi-kde

SOURCES += wtf.cpp

I didn't achieve what I wanted, but I feel like I achieved something magical and beautiful after all.

I shall now perform some haruspicy on those oscure cmake files to see if I can figure something out. But seriously, people?

Posted Mon Feb 23 11:36:18 2015 Tags:

Setting up Akonadi

Now that I have a CalDAV server that syncs with my phone I would like to use it from my desktop.

It looks like akonadi is able to sync with CalDAV servers, so I'm giving it a try.

First thing first is to give a meaning to the arbitrary name of this thing. Wikipedia says it is the oracle goddess of justice in Ghana. That still does not hint at all at personal information servers, but seems quite nice. Ok. I gave up with software having purpose-related names ages ago.

# apt-get install akonadi-server akonadi-backend-postgresql

Akonadi wants a SQL database as a backend. By default it uses MySQL, but I had enough of MySQL ages ago.

I tried SQLite but the performance with it is terrible. Terrible as in, it takes 2 minutes between adding a calendar entry and having it show up in the calendar. I'm fascinated by how Akonadi manages to use SQLite so badly, but since I currently just want to get a job done, next in line is PostgreSQL:

# su - postgres
$ createuser enrico
$ psql postgres
postgres=# alter user enrico createdb;

Then as enrico:

$ createdb akonadi-enrico
$ cat <<EOT > ~/.config/akonadi/akonadiserverrc
[%General]
Driver=QPSQL

[QPSQL]
Name=akonadi-enrico
StartServer=false
Host=
Options=
ServerPath=
InitDbPath=

I can now use kontact to connect Akonadi to my CalDAV server and it works nicely, both with calendar and with addressbook entries.

KDE has at least two clients for Akonadi: Kontact, which is a kitchen sink application similar to Evolution, and KOrganizer, which is just the calendar and scheduling component of Kontact.

Both work decently, and KOrganizer has a pretty decent startup time. I now have a usable desktop PIM application that is synced with my phone. W00T!

Next step is to port my swift little calendar display tool to use Akonadi as a back-end.

Posted Tue Feb 17 15:34:55 2015 Tags:

seat-inspect

Four months ago I wrote this somewhere:

Seeing a DD saying "this new dbus stuff scares me" would make most debian users scared. Seeing a DD who has an idea of what is going on, and who can explain it, would be an interesting and exciting experience.

So, let's be exemplary, competent and patient. Or at least, competent. Some may like or not like the changes, but do we all understand what is going on? Will we all be able to support our friends and customers running jessie?

I confess that although I understand the need for it, I don't feel competent enough to support systemd-based machines right now.

So, are we maybe in need of help, cheat sheets, arsenals of one-liners, diagnostic tools?

Maybe a round of posts on -planet like "one debian package a day" but with new features that jessie will have, and how to understand them and take advantage of them?

That was four months ago. In the meantime, I did some work, and it got better for me.

Yesterday, however, I've seen an experienced Linux person frustrated because the shutdown function of the desktop was doing nothing whatsoever. Today I found John Goerzen's post on planet.

I felt like some more diagnostic tools were needed, so I spent the day making seat-inspect.

seat-inspect tries to make the status of the login/seat system visible, to help with understanding and troubleshooting.

The intent of running the code is to have an overview of the system status, both to see what the new facilities are about, and to figure out if there is something out of place.

The intent of reading the code is to have an idea of how to use these facilities: the code has been written to be straightforward and is annotated with relevant bits from the logind API documentation.

seat-inspect is not a finished tool, but a starting point. I put it on github hoping that people will fork it and add their own extra sanity checks and warnings, so that it can grow into a standard thing to run if a system acts weird.

As it is now, it should be able to issue warnings if some bits are missing for network-manager or shutdown functions to work correctly. I haven't really tested that, though, because I don't have a system at hand where they are currently not working fine.

Another nice thing of it is that when running seat-inspect -v you get a dump of what logind/consolekit think about your system. I found it an interesting way to explore the new functionalities that we recently grew. The same can be done, and in more details, with loginctl calls, but I lacked a summary.

After writing this I feel a bit more competent, probably enough to sit at somebody's computer and poke into loginctl bits. I highly recommend the experience.

Posted Tue Feb 10 18:06:43 2015 Tags:
Posted Sat Jun 6 00:57:39 2009

Pages about OpenMoko.

Released nodm 0.7

I have released version 0.7 of nodm.

It only fixes one silly typo in autotools, which made it fail to build on Fedora.

Posted Sun May 23 21:36:52 2010 Tags:

Released nodm 0.6

I have released version 0.6 of nodm.

It is purely a bug fix release, trying harder to detect a console in order to get rid of a bug introduced with version 0.5

Posted Mon Aug 3 12:34:16 2009 Tags:

Released nodm 0.5

I have released version 0.5 of nodm.

New features:

  • truncate ~/.xsession-errors on startup: finally that file stops growing, and growing, and growing...
  • dynamic VT allocation: it can now avoid opening a virtual terminal if it is already in use.
Posted Fri Jul 24 02:29:55 2009 Tags:

Getting dbus signatures right from Vala

I am trying to play a bit with Vala on the FreeRunner.

The freesmartphone.org stack on the OpenMoko is heavily based on DBus. Using DBus from Vala is rather simple, if mostly undocumented: you get a few examples in the Vala wiki and you make do with those.

All works fine with simple methods. But what with providing callbacks to signals that have complex nested structures in their signatures, like aa{sv}? You try, and then if you don't get the method signature right, the signal is just silently not delivered because it does not match the method signature.

So this is how to provide a callback to org.freesmartphone.Usage.ResourceChanged, with signature sba{sv}:

public void on_resourcechanged(dynamic DBus.Object pos,
                   string name,
                   bool state,
                   HashTable<string, Value?> attributes)
{
    stderr.printf("Resource %s changed\n", name);
}

And this is how to provide a callback to org.freesmartphone.GPS.UBX.DebugPacket, with signature siaa{sv}:

protected void on_ubxdebug_packet(dynamic DBus.Object ubx, string clid, int length,
        HashTable<string, Value?>[] wrongdata)
{
    stderr.printf("Received UBX debug packet");

    // Ugly ugly work-around
    PtrArray< HashTable<string, Value?> >* data = (PtrArray< HashTable<string, Value?> >)wrongdata;

    stderr.printf("%u elements received", data->len);
}

What is happening here is that the only method signature that I found matching the dbus signature is this one. However, the unmarshaller for some reason gets it wrong, and passes a PtrArray instead of a HashTable array. So you need to cast it back to what you've actually been passed.

Figuring all this out took several long hours and was definitely not fun.

Posted Wed Jul 15 12:30:50 2009 Tags:

Mapping using the Openmoko FreeRunner headset

The FreeRunner has a headset which includes a microphone and a button. When doing OpenStreetMap mapping, it would be very useful to be able to keep tangogps on the display and be able to mark waypoints using the headset button, and to record an audio track using the headset microphone.

In this way, I can use tangogps to see where I need to go, where it's already mapped and where it isn't, and then I can use the headset to mark waypoints corresponding to the audio track, so that later I can take advantage of JOSM's audio mapping features.

Enter audiomap:

$ audiomap --help
Usage: audiomap [options]

Create a GPX and audio trackFind the times in the wav file when there is clear
voice among the noise

Options:
  --version      show program's version number and exit
  -h, --help     show this help message and exit
  -v, --verbose  verbose mode
  -m, --monitor  only keep the GPS on and monitor satellite status
  -l, --levels   only show input levels

If called without parameters, or with -v which is suggested, it will:

  1. Fix the mixer settings so that it can record from the headset and detect headset button presses.
  2. Show a monitor of GPS satellite information until it gets a fix.
  3. Synchronize the system time with the GPS time so that the timestamps of the files that are created afterwards are accurate.
  4. Start recording a GPX track.
  5. Start recording audio.
  6. Record a GPX waypoint for every headset button press.

When you are done, you stop audiomap with ^C and it will properly close the .wav file, close the tags in the GPX waypoint and track files and restore the mixer settings.

You can plug the headset out and record using the handset microphone, but then you will not be able to set waypoints until you plug the headset back in.

After you stop audiomap, you will have a track, waypoints and .wav file ready to be loaded in JOSM.

Big thanks go to Luca Capello for finding out how to detect headset button presses.

Posted Sun Jun 7 23:51:37 2009 Tags:

Simple tool to query the GPS using the OpenMoko FSO stack

I was missing a simple command line tool that allows me to perform basic GPS queries in shellscripts.

Enter getgps:

# getgps --help
Usage: getgps [options]

Simple GPS query tool for the FSO stack

Options:
  --version          show program's version number and exit
  -h, --help         show this help message and exit
  -v, --verbose      verbose mode
  -q, --quiet        suppress normal output
  --fix              check if we have a fix
  -s, --sync-time    set system time from GPS time
  --info             get all GPS information
  --info-connection  get GPS connection information
  --info-fix         get GPS fix information
  --info-position    get GPS position information
  --info-accuracy    get GPS accuracy information
  --info-course      get GPS course information
  --info-time        get GPS time information
  --info-satellite   get GPS satellite information

So finally I can write little GPS-aware scripts:

if getgps --fix -q
then
    start_gps_aware_program
else
    start_gps_normal_program
fi

Or this.

Posted Sun Jun 7 17:59:32 2009 Tags:

Voice-controlled waypoints

I have it in my TODO list to implement taking waypoints when pressing the headset button of the openmoko, but that is not done yet.

In the meantime, I did some experiments with audio mapping, and since I did not manage to enter waypoints while recording them, I was looking for a way to make use of them anyway.

Enter findvoice:

$ ./findvoice  --help
Usage: findvoice [options] wavfile

Find the times in the wav file when there is clear voice among the noise

Options:
  --version             show program's version number and exit
  -h, --help            show this help message and exit
  -v, --verbose         verbose mode
  -p NUM, --percentile=NUM
            percentile to use to discriminate noise from voice
            (default: 90)
  -t, --timestamps      print timestamps instead of human readable information

You give it a wav file, and it will output a list of timestamps corresponding to where it things that you were talking clearly and near the FreeRunner / voice recorder instead of leaving the recorder dangling to pick up background noise.

Its algorithm is crude and improvised because I have no background whatsoever in audio processing, but it basically finds those parts of the audio file where the variance of the samples is above a given percentile: the higher the percentile, the less timestamps you get; the lower the percentile, the more likely it is that it picks a period of louder noise.

For example, you can automatically extract waypoints out of an audio file by using it together with gpxinterpolate:

./findvoice -t today.wav | ./gpxinterpolate today.gpx > today-waypoints.gpx

The timestamps it outputs are computed using the modification time of the .wav file: if your system clock was decently synchronised (which you can do with getgps), then the mtime of the wav is the time of the end of the recording, which gives the needed reference to compute timestamps that are absolute in time.

For example:

getgps --sync-time
arecord file.wav
^C
./findvoice -t file.wav | ./gpxinterpolate today.gpx > today-waypoints.gpx
Posted Sun Jun 7 02:48:40 2009 Tags:

Geocoding Unix timestamps

Geocoding EXIF tags in JPEG images is fun, but there is more that can benefit from interpolating timestamps over a GPX track.

Enter gpxinterpolate:

$ ./gpxinterpolate --help
Usage: gpxinterpolate [options] gpxfile [gpxfile...]

Read one or more GPX files and a list of timestamps on standard input. Output
a GPX file with waypoints at the location of the GPX track at the given
timestamps.

Options:
  --version      show program's version number and exit
  -h, --help     show this help message and exit
  -v, --verbose  verbose mode

For example, you can create waypoints interpolating file modification times:

find . -printf "%Ts %p\n" | ./gpxinterpolate ~/tracks/*.gpx > myfiles.gpx

In case you wonder where you were when you modified or accessed a file, now you can find out.

Posted Sun Jun 7 02:07:43 2009 Tags:

Recording audio on the FreeRunner

The FreeRunner can record audio. It is nice to record audio: for example I can run the recording in background while I keep tangogps in the screen, and take audio notes about where I am while I am doing mapping for OpenStreetMap.

Here is the script that I put together to create geocoded audio notes:

#!/bin/sh

WORKDIR=~/rec
TMPINFO=`mktemp $WORKDIR/info.XXXXXXXX`

# Sync system time and get GPS info
echo "Synchronising system time..."
getgps --sync-time --info > $TMPINFO

# Compute an accurate basename for the files we generate
BASENAME=~/rec/rec-$(date +%Y-%m-%d-%H-%M-%S)
# Then give a proper name to the file with saved info
mv $TMPINFO $BASENAME.info

# Proper mixer settings for recording
echo "Recording..."
alsactl -f /usr/share/openmoko/scenarios/voip-handset.state restore
arecord -D hw -f cd -r 8000 -t wav $BASENAME.wav

echo "Done"

It works like this:

  1. It synchronizes the system time from the GPS (if there is a fix) so that the timestamps on the wav files will be as accurate as possible.
  2. It also gets all sort of information from the GPS and stores them into a file, should you want to inspect it later.
  3. It records audio until it gets interrupted.

The file name of the files that it generates corresponds to the beginning of the recording. The mtime of the wav file obviously corresponds to the end of the recording. This can be used to later georeference the start and end point of the recording.

You can use this to check mixer levels and that you're actually getting any input:

arecord -D hw -f cd -r 8000 -t wav -V mono /dev/null

The getgps script is now described in its own post.

You may now want to experiment, in JOSM, with "Preferences / Audio settings / Modified times (time stamps) of audio files".

Posted Sun Jun 7 01:30:37 2009 Tags:

Polysms

Here is my first software designed for the FreeRunner: polysms. It's a commandline tool: you pass it a polygen grammar name and a phone number, and it will send a SMS to that phone number using the polygen output for that grammar as the SMS text:

# polyrun manager 0012345678

And here is the code, that works on the http://www.freesmartphone.org dbus framework:

#!/usr/bin/python

# (C) 2008 Enrico Zini
# Most bits of this are stripped from zhone, which is:
# (C) 2007 Johannes 'Josch' Schauer
# (C) 2008 Michael 'Mickey' Lauer <mlauer@vanille-media.de>
# (C) 2008 Jan 'Shoragan' Luebbe
# (C) 2008 Daniel 'Alphaone' Willmann
# (C) 2008 Openmoko, Inc.
# GPLv2 or later

from dbus import SystemBus, Interface
from dbus.exceptions import DBusException
import logging
logger = logging.getLogger( __name__ )

from dbus.mainloop.glib import DBusGMainLoop
DBusGMainLoop(set_as_default=True)

import gobject
import sys
from subprocess import Popen, PIPE


class Phone:
    def tryGetProxy( self, busname, objname ):
        try:
            return self.bus.get_object( busname, objname )
        except DBusException, e:
            logger.warning( "could not create proxy for %s:%s" % ( busname, objname ) )

    def __init__(self):
        try:
            self.bus = SystemBus()
        except DBusException, e:
            logger.error( "could not connect to dbus_object system bus: %s" % e )
            return False

        # Phone
        self.gsm_device_obj = self.tryGetProxy( 'org.freesmartphone.ogsmd', '/org/freesmartphone/GSM/Device' )

        if ( self.gsm_device_obj is not None ):
            self.gsm_device_iface = Interface(self.gsm_device_obj, 'org.freesmartphone.GSM.Device')
            self.gsm_sim_iface = Interface(self.gsm_device_obj, 'org.freesmartphone.GSM.SIM')
            self.gsm_network_iface = Interface(self.gsm_device_obj, 'org.freesmartphone.GSM.Network')
            self.gsm_call_iface = Interface(self.gsm_device_obj, 'org.freesmartphone.GSM.Call')
            self.gsm_test_iface = Interface(self.gsm_device_obj, 'org.freesmartphone.GSM.Test')

        # Main loop
        self.loop = gobject.MainLoop()

    def send(self, number, message):
        def onSent():
            print "SENT"
            self.loop.quit()
        def onStore(index):
            print "STORED AS", index
            self.gsm_sim_iface.SendStoredMessage(
                index,
                reply_handler=onSent,
                error_handler=self.onError
            )
        self.gsm_sim_iface.StoreMessage(
            number, message,
            reply_handler=onStore,
            error_handler=self.onError
        )

    def onError(self, result):
        print "ERROR", result

    def mainloop(self):
        self.loop.run()

if len(sys.argv) != 3:
    print >>sys.stderr, "Usage: %s grammarname phonenumber"
    sys.exit(1)

message = Popen(["/usr/bin/polyrun", sys.argv[1]], stdout=PIPE).communicate()[0]
number = sys.argv[2]

print "Sending to %s:" % number
print message

phone = Phone()
phone.send(number, message)
phone.mainloop()
Posted Sat Jun 6 00:57:39 2009 Tags:
Posted Sat Jun 6 00:57:39 2009

Pages about Debtags.

Evolution's old odd mail folders to mbox

Something wrong happened in my dad's Evolution. It just would get stuck checking mail forever, with no useful diagnostic that I could find. Fun. Not.

Anyway, I solved by resetting everything to factory defaults, moving away all gconf entries and .evolution/ files. Then it started to work again, of course then I needed to reconfigure it from scratch.

It turned out however that some old mail was only archived locally, and in a kind of weird format that looks like this:

$ ls -la Enrico/
total 336
drwx------ 2 enrico enrico   4096 Jul 23 03:05 .
drwxr-xr-x 7 enrico enrico   4096 Jul 23 03:12 ..
-rw------- 1 enrico enrico   3230 Dec  4  2010 113.HEADER
-rw------- 1 enrico enrico  14521 Dec  4  2010 113.TEXT
-rw------- 1 enrico enrico   3209 Oct 22  2010 134.HEADER
-rw------- 1 enrico enrico   2937 Oct 22  2010 134.TEXT
-rw------- 1 enrico enrico   3116 Jun 27  2011 15.
-rw------- 1 enrico enrico   3678 Jun 27  2011 168.
-rw------- 1 enrico enrico     73 Apr 27  2009 22.1.MIME
-rw------- 1 enrico enrico   3199 Apr 27  2009 22.2
-rw------- 1 enrico enrico     88 Apr 27  2009 22.2.MIME
[...]

I couldn't even find the name of that mail folder layout, let alone conversion tools. So I had to sit down and waste my sunday break writing software to convert that to a mbox file. Here's the tool, may it save you the awful time I had today: http://anonscm.debian.org/gitweb/?p=users/enrico/evo2mbox.git

Note: feel free to fork it, or send patches, but don't bother with feature requests. Evolution isn't and won't be a personal interest of mine. Anything that makes an afternoon at my parents more tiresome than a whole busy month of paid work, doesn't deserve to be.

Luckily they now seem to have changed the local folder format to Maildir.

Posted Mon Jul 23 03:27:50 2012 Tags:

Giving away distromatch

at last year's Fosdem I tried to inject a lot of energy into distromatch but shortly afterwards I've had to urgently rewrite the nm.debian.org website.

After Lars Wirzenius GTDFH talks in Bologna and Varese I wrote a tool which, among other things, is able to scan my home dir and list how many projects I'm working on.

The output was scary. Like, they are too many. Like, I couldn't even recite the list out of memory. And since I couldn't do that, I had no idea there were so many. And I kept being stressful because I couldn't manage to take care of them all properly.

Now that I became conscious of the situation, it's time to deal with it like a grown up, and politely back off from some of my irresponsible responsibilities.

Distromatch is one of them. It had just started as a proof of concept prototype, and I had the vision that it could be the basis for a fantastic culture of sharing and exchange of information across distributions.

I need to distinguish the vision from the responsibility. I still have that vision for distromatch, but I cannot take responsibility for making it happen.

So I am giving it up to anyone who has the time and resources to pick up that responsibility.

Current status

It works well enough as a prototype. I believe it can successfully map a large enough slice of packages, that one can prototype stuff based on it.

I have for example used it to export the Debtags categories for other distros, and the resulting file looked big enough to be used for prototyping category-based features on distributions that don't have them yet.

I think it also works well enough to support a few common use cases, like sharing screenshots, or doing most of the work of converting dependency lists from a distro to another.

And finally, anyone can deploy it, and work on it.

Existing data sources

Everything I index in the Debian distromatch deployment is available at http://dde.debian.net/exports/distromatch/. The rpm-based data in there comes from an export script I wrote that runs on Sophie, but which I cannot maintain properly.

This is an experimental export of Fedora and OpenSUSE data: http://tmp.vuntz.net/misc/distromatch/distromatch-opensuse-fedora.tar

All existing export scripts are found in distromatch git repo on gitorious.

Contacts I gathered at Fosdem

At Fosdem I devoted quite some work to get contacts from all possible distributions and software repositories, so that distromatch could be hooked into them. Here is a dump of what I have collected:

  • Debian: me
  • OpenSuse: Vincent Untz and Adrian Schröter
  • Fedora: Tom "Spot" Callaway
  • Arch: Tasser on IRC
  • CPAN: contact the people of https://metacpan.org/, on irc.perl.org:#metacpan or make an issue on github
  • NetBSD: ask on #netbsd on Freenode
  • FreeBSD: Baptiste Daroussin (bapt)
  • Mageia: Olivier Thauvin

Some of those contacts may have "expired" in the meantime: I wouldn't assume all of them still remember talking with me, although most probably still do.

My commitment for the time being

I am happy to commit, at the moment, to maintaining a working data export for Debian data. I can take responsibility for making it so that the Debian data for it stays up to date, and to fix it asap if it isn't the case.

I hope that now someone can take distromatch over from me, and make it grow to achieve its great potential.

Posted Sat Jul 21 16:54:18 2012 Tags:

More diversity in Debian skills

This blog post has been co-authored with Francesca Ciceri.

In his Debconf talk, zack said:

We need to understand how to invite people with different backgrounds than packaging to join the Debian project [...] I don't know what exactly, but we need to do more to attract those kinds of people.

Francesca and I know what we could do: make other kinds of contributions visible.

Basically, we should track and acknowledge the contributions of webmasters, translators, programmers, sysadmins, event organisers, and so on, at the same level as what we do for packagers: DDPO, minechangelogs, Portfolio...

For any non-packaging activity that we can make visible and credited, we get:

  • to acknowledge the people who do it, and show that they are active contributors in the project;

  • to acknowledge the work that gets done, and show the actual amount of non-packaging work that gets done in Debian every day;

  • to allow non-packagers to have a reputation, too: first of all, they deserve it, and among other things, it would make nm processing trivial.  

Here's an example: who's the lead translator for German? And if you are German, who's the lead translator for Spanish? Czech? Thai? I (Enrico) don't know the answers, not even for Italian, but we all should! Or at least it should be trivial to find out.

To start to change this, is just a matter of programming.

Francesca already worked on a list of trackable data sources, at least for translators.

Here are some more details, related to translation:

  • Translations can be tracked via the i18n robot (and relative statistics). This works only with teams who activated the robot and actively use the pseudo-urls in their messages on localisation mailing lists. Some translators don't bother to do it but it's ok to only support the main workflow. It beats extracting .po files from l10n-tagged BTS bugs at any rate.

  • DPN and website translations: for wml pages there's a specific field to be extracted for each translated page: grep for maintainer="name" on normal wml pages, while for DPN translations we have a specific translator="name" field. The problem is that this field is not mandatory, so sometimes there's no indication of the maintainer. Again, it's ok to only support the main workflow.

    Anyway, this is preferable to the cvs log: often the commit is done by the coordinator of the team and not by the actual translator. See above for the alternative solution of using the statistics provided by the i18n bot.

  • DDTSS: since the new release of DDTSS-Django, done by Martijn van Oosterhout about a year ago, the contributions are by default non-anonymous. This should be easy to track.

  • http://wiki.debian.org: it is more complicated because in the wiki we do not have a proper l10n translation workflow, so the only thing that can be tracked are changelogs $LANG/* pages. A nice idea would be to have translated pages list the version of the page that was translated and who did the translation.

  • translation of debian manuals and release notes: usually in the translation of manuals and long documentation there is a specific translator field.

And here are some notes about other fields:

  • DPN editors: for each issue there's a list of editors at the bottom of the page. In the wml: grep for editor=.

  • Artwork: artwork submitted via debianart are easy to track on the portal. Anyway usually you can find the author in the license and copyright file.

  • Programming: the only thing we have is the list of services which can be expanded if needed.

  • Press and publicity: there seems to be not much besides svn logs.

  • l10n-english: The Smith Review Project page has some tracking links. Other activities can probably only be tracked, at the moment, via mailing list activity.

  • Events: we can use the "main coordinator" field on www.debian.org/events/$year/$date-$eventname.wml: grep for <define-tag coord>; for events not published on the http://www.debian.org, but only on http://wiki.debian.org, the coordinator or the contact for the event is usually present on the page itself.

  • Sysadmins: we haven't asked DSA.

And finally, if you are still wondering who those translation coordinators are, they are listed here, although not all teams keep that page up to date.

Of course, when a data source is too hard to mine, it can make sense to see if the workflow could be improved, rather than spending months writing compicated mining code.

This is a fun project for people at Debconf to get together and try.

If by the end of the conference we had a way to credit some group of non-packaging contributors, even if just one like translators or website contributors, at least we would finally have started having official trackers for the activities of non-packagers.

Posted Thu Jul 12 14:01:54 2012 Tags:

Debtags for derivative distributions

Sometimes I do cool stuff and I forget to announce it.

Ok, so I recently announced a new Debtags website.

I forgot to say in the announcement that the new website does not only know of Debian packages: see for example this page, at the very bottom it says: "Distributions: oneiric, precise, sid, testing".

This means that already, here and now, debtags.debian.net can be used to tag packages from both Debian and Ubuntu, and can easily be extended to cover the entire Debian ecosystem.

If you are a package maintainer, you will notice that your maintainer page shows your packages from everywhere. If you want to filter things a bit, for example hide obsolete packages from an old Debian Stable or Ubuntu LTS, just click on the "Settings" link on the top right to configure the page.

How it works

The magic is in this mergepackages script, which is run daily, and exports merged Packages files at dde.debian.net. The debtags.debian.net concept of Packages and Sources files are just those all-merged.gz and all-merged-sources.gz.

The merging is simple: that rebuild script processes files in order, and the first version of a package that is found is chosen as the base for the one that will go in the merged Packages file. Some fields like "Description" are just taken from this pivot package, others like Architecture or dependencies are merged into it. It's arbitrary, but works for me: the result has all the packages with all their possible architectures and dependencies, and is ready to be indexed with apt-xapian-index.

At the moment I pull data from Debian and Ubuntu, but you can see that the script can easily be extended to pull data from any Debian-style ftp archive, so any Debian derivative can go in. I've already started negotiations with the Derivatives Census on how to add any Debian derivative and keep the list up to date.

How to export tags for your own distribution

I'll use Ubuntu as an example since the data is already available.

The way you add Debtags to the Ubuntu packages file is just this one:

  1. Get the full reviewed tag database
  2. Optionally filter out those packages that you are not interested in
  3. Tweak this script to build an overrides file.
  4. Give the overrides file to your favourite ftp archive building tool.

The make-overrides is a bit rusty: if you improve it, please send me your changes.

That is it, nothing else required, no excuses, it's ready, here, now!

Hitches and gotchas

This merged Packages file is a bit of a hack, and suffers from name conflicts across distributions, where two different softwares are packaged in two different distributions with the same name.

Ideally, name conflicts should not happen: if a derivative decided to package kate and call it gedit, they deserve to have it tagged uitoolkit::gtk. I think it's rather important that the whole Debian ecosystem works as much as possible with a single package namespace.

However, that reasoning fails if you take time into account: packages get renamed, like git and chromium, and may mean completely different things, for example, if you compare Debian Stable with Debian Sid.

This last is a problem caused by debtags only working with package names but not package versions. I have a strategy in mind based on being able to override the stable tag database using headers in debian/control; it still needs some details sorted out, but I'm confident we will be able to address these issues properly soon enough.

Why stop at the Debian ecosystem?

Why indeed. I'm clearly trying to use FOSDEM, and the CrossDistribution devroom as the venue to discuss just that.

Posted Fri Jan 20 15:12:33 2012 Tags:

Deploying distromatch

I have been working on allowing anyone to set up their own distromatch instance.

For Debian and Ubuntu, I can easily generate the distromatch input using UDD and the Contents files found in any mirrors.

For the whole RPM world, thanks to Olivier Thauvin I have been able to set up regular exports from the vast Sophie database.

I have set up distromatch access on DDE, which can also serve as a list of all working distributions so far. If you have access to the full dataset of package names and package contents for a distribution not in that list, please get in touch and we can add it.

I'm also exporting the full raw dataset which enables anyone to set up the same distromatch environment on their own machines.

Here is how:

# Get distromatch
git clone git://gitorious.org/appstream/distromatch.git
cd distromatch

# Fetch distribution information (updated every 2 days)
wget http://dde.debian.net/exports/distromatch-all.tar.gz

# Unpack it
mkdir data
tar -C data -zxf distromatch-all.tar.gz

# Reindex it (use --verbose if you are curious)
./distromatch --datadir=data --reindex --verbose

# Run it
./distromatch --datadir=data debian gedit

What does this mean? For example it means that if another distribution has some data (categories, screenshots...) that your distribution doesn't have, you can use distromatch to translate package names, then go and get it!

My next step is going to be to improve the distromatch functionality in DDE and possibly build a simple user friendly web interface to it. If you have some JQuery experience and would like to help, don't wait to get in touch.

Posted Fri Feb 18 13:46:30 2011 Tags:

update-apt-xapian-index on other distros

I've drafted a little HOWTO on using apt-xapian-index on non-Debian distributions.

The procedure has been tried on Mageia with some success, and there's no reason it wouldn't work everywhere else: the index itself does not depend on anything distro-specific.

Posted Tue Jan 25 23:01:45 2011 Tags:

A prototype webby markety appy thing

What better way to introduce my work at an Application Installer meeting than to come with a prototype package browser modeled after shopping sites developed in just a few hours?

It's a little Flask webapp that just works on any Debian system, using the local apt-xapian-index as a backend. It has fast keyword search, faceted navigation and screenshots, and it runs on your system showing the packages that you have available.

Screenshot of packageshelf

To try it:

git clone git://git.debian.org/users/enrico/pkgshelf.git
cd pkgshelf
./web-server.py

Then visit http://localhost:5000

It didn't have much interface polishing, as it's just a quick technology demo. However you can see that:

  • keyword search is fast (fast enought that it could be made to search as you type);
  • relevant tags appear on the left, grouped by facets;
  • the most relevant tags are highlighted;
  • the less relevant tags could be hidden behind a [more] expander;
  • you can choose several strategies to hide packages you may find irrelevant.

Things that need doing:

  • hiding uninteresting facets;
  • making it pretty.

It's essentially JavaScript and CSS work. Anyone wants to play?

Posted Sat Jan 22 01:40:50 2011 Tags:

Match package names across distributions

What would happen if we had a quick and reliable way to match package names across distributions?

These ideas came up at the appinstaller2011 meeting:

  • it would be easy to lookup screenshots in the local distro, and if there are none then fall back on other distributions;
  • it would be easy to port Debtags to other distributions, and possibly get changes back;
  • it would be trivial to add a [patches in $DISTRO] link to the PTS
  • it would be easy to point to other BTSes

We thought they were good ideas, so we started hacking.

To try it, you need to get the code and build the index first:

git clone git://git.debian.org/users/enrico/distromatch.git
cd distromatch
# Careful: 90Mb
wget http://people.debian.org/~enrico/dist-info.tar.gz
tar zxf dist-info.tar.gz
# Takes a long time to do the indexing
./distromatch --reindex --verbose

Then you can query it this way:

./distromatch $DISTRO $PKGNAME [$PKGNAME1 ...]

This would give you, for the package $PKGNAME in $DISTRO, the corresponding package names in all other distros for which we have data. If you do not provide package names, it automatically shows output for all packages in $DISTRO.

For example:

$ time ./distromatch debian libdigest-sha1-perl
debian:libdigest-sha1-perl fedora:perl-Digest-SHA1
debian:libdigest-sha1-perl mandriva:perl-Digest-SHA1
debian:libdigest-sha1-perl suse:perl-Digest-SHA1

real    0m0.073s
user    0m0.056s
sys 0m0.016s

Yes it's quick. It builds a Xapian index with the information it needs, and then it reuses it. As soon as I find a moment, I intend to deploy an instance of it on DDE.

It is using a range of different heuristics:

  • match packages by name;
  • match packages by desktop files contained within;
  • match packages by pkg-config metadata files contained within;
  • match packages by [/usr]/bin/* files contained within;
  • match packages by shared library files contained within;
  • match packages by devel library files contained within;
  • match packages by man pages contained within;
  • match stemmed form of development library package names;
  • match stemmed form of shared library package names;
  • match stemmed form of perl library package names;
  • match stemmed form of python library package names.

This list may get obsolete soon as more heuristics get implemented.

Euristics will never cover all corner cases we surely have, but the idea is that if we can match a sizable amout of packages, the rest can be somehow fixed by hand as needed.

The data it requires for a distribution should be rather straightforward to generate:

  1. a file which maps binary package names to source package names
  2. a file with the list of files in all the packages

For example:

$ ls -l dist-debian/
total 39688
-rw-r--r--  1 enrico enrico  1688249 Jan 20 17:37 binsrc
drwxr-xr-x  2 enrico enrico     4096 Jan 21 19:12 db
-rw-r--r--  1 enrico enrico 29960406 Jan 21 10:02 files.gz
-rw-r--r--  1 enrico enrico  8914771 Jan 21 18:39 interesting-files

$ head dist-debian/binsrc 
openoffice.org-dev openoffice.org
ext4-modules-2.6.32-5-4kc-malta-di linux-kernel-di-mipsel-2.6
linux-headers-2.6.30-2-common linux-2.6
libnspr4 nspr
ipfm ipfm
libforks-perl libforks-perl
med-physics debian-med
libntfs-3g-dev ntfs-3g
libguppi16 guppi
selinux selinux

$ zcat dist-debian/files.gz | head
memstat etc/memstat.conf
memstat usr/bin/memstat
memstat usr/share/doc/memstat/changelog.gz
memstat usr/share/doc/memstat/copyright
memstat usr/share/doc/memstat/memstat-tutorial.txt.gz
memstat usr/share/man/man1/memstat.1.gz
libdirectfb-dev usr/bin/directfb-config
libdirectfb-dev usr/bin/directfb-csource
libdirectfb-dev usr/include/directfb-internal/core/clipboard.h
libdirectfb-dev usr/include/directfb-internal/core/colorhash.h

interesting-files and db are generated when indexing.

To prove the usefulness of the idea (but does it need proving?), you can find in the same git repo a little example app (it took me 10 minutes to write it), that uses the distromatch engine to export Debtags tags to other distributions:

$ ./exportdebtags fedora | head
memstat: admin::benchmarking, interface::commandline, role::program, use::monitor
libdirectfb-dev: devel::lang:c, devel::library, implemented-in::c, interface::framebuffer, role::devel-lib
libkonqsidebarplugin4a: implemented-in::c++, role::shared-lib, suite::kde, uitoolkit::qt
libemail-simple-perl: devel::lang:perl, devel::library, implemented-in::perl, role::devel-lib, role::shared-lib, works-with::mail
libpoe-component-pluggable-perl: devel::lang:perl, devel::library, implemented-in::perl, role::shared-lib
manpages-ja: culture::japanese, made-of::man, role::documentation
libhippocanvas-dev: devel::library, qa::low-popcon, role::devel-lib
libexpat-ocaml-dev: devel::lang:ocaml, devel::library, implemented-in::c, implemented-in::ocaml, role::devel-lib, works-with-format::xml
libgnutls-dev: devel::library, role::devel-lib, suite::gnu

Just in case this made you itch to play with Debtags in a non-Debian distribution, I've generated the full datasets for Fedora, Mandriva and OpenSUSE.

Others have been working on the same matching problem. After we started writing code we started to become aware of existing work:

I'd like to make use of those efforts, maybe to cross-validate results, maybe even better as yet another heuristics.

Update:

I built a simple distromatch query system into DDE!

Posted Sat Jan 22 01:40:50 2011 Tags:

Cross-distro Meeting on Application Installer

I have been to a Cross-distro Meeting on Application Installer which to the best of our knowledge is also the first one of its kind. Credit goes to Vincent Untz for organising it, to OpenSUSE for hosting it and to the various sponsors for getting us there.

It went surprisingly well. We got along, got stuff done, did as much work as possible to agree on as many formats, protocols and technologies as we possibly could.

The timing of it is very important, as most major distros would like to adopt some of the features that just became popular in the various new app markets and stores, such as screenshots, user comments and ratings. It looks like a lot of new code is about to be written, or a lot of existing code is about to gain quite a bit of popularity.

For my part, I presented the work on Debtags and apt-xapian-index.

With regards to Debtags, other distros seem to be missing a compehensive classification system, and Debtags is, well, it.

With regards to apt-xapian-index, we just noticed that it's the perfect back-end for what everyone would like to do, and the index structure is rather distribution-agnostic, and it's been road-tested with considerable success by at least software-center, so it attracted quite a bit of interest, and will likely attract some more.

Just to prove a point I put together a prototype webby markety appy thing in just a few hours of work.

The meeting was also the ideal place to create a joint effort to match package names across distributions, which means that a lot of things that were hard to share before, such as screenshots, tags and patches, are suddenly not hard to share anymore.

Posted Sat Jan 22 01:40:50 2011 Tags:

fuss-launcher: an application launcher built on apt-xapian-index

Long ago I blogged about using apt-xapian-index to write an application launcher.

Now I just added a couple of new apt-xapian-index plugins that look like they have been made just for that.

In fact, they have indeed been made just for that.

After my blog post in 2008, people from Truelite and the FUSS project took up the challenge and wrote a launcher applet around my example engine.

The prototype has been quite successful in FUSS, and as a consequence I've been asked (and paid) to bring in some improvements.

The result, that I have just uploaded to NEW, is a package called fuss-launcher:

* New upstream release
   - Use newer apt-xapian-index: removed need of local index
   - Dragging a file in the launcher shows the applications that can open it
   - Remembers the applications launched more frequently
   - Allow to set a list of favourite applications

To get it:

  • apt-get install fuss-launcher (after it passed NEW);
  • or git clone http://git.fuss.bz.it/git/launcher.git/ and apt-get install python-gtk2 python-xapian python-xdg apt-xapian-index app-install-data

It requires apt-xapian-index >= 0.35.

To try it:

  1. Make sure your index is up to date, especially if you just installed app-install-data: just run update-apt-xapian-index as root.
  2. Run fuss-launcher.
  3. Click on the new tray icon to open the launcher dialog.
  4. Type some keywords and see the list of matching applications come to life as you type.

It's worth mentioning again that all this work was sponsored by Truelite and the Fuss project, which rocks.

Some screenshots:

When you open the launcher, by default it shows the most frequently started applicationss and the favourite applications:

launcher just opened

When you type some keywords, you get results as you type, and context-sensitive completion:

keyword search

When you drag a file on the launcher you only see the applications that can open that file:

drag files to the launcher

Posted Mon May 17 10:41:09 2010 Tags:
Posted Sat Jun 6 00:57:39 2009
sw

Software

Work around Google evil .ics feeds

I've happily been using 2015/akonadi-install for my calendars, and yesterday I added an .ics feed export from Google, as a URL file source. It is a link in the form: https://www.google.com/calendar/ical/person%40gmail.com/private-12341234123412341234123412341234/basic.ics

After doing that, I noticed that the fan in my laptop was on more often than usual, and I noticed that akonadi-server and postgres were running very often, and doing quite a lot of processing.

The evil

I investigated and realised that Google seems to be doing everything they can to make their ical feeds hard to sync against efficiently. This is the list of what I have observed Gmail doing to an unchanged ical feed:

  • Date: headers in HTTP replies are always now
  • If-Modified-Since: is not supported
  • DTSTAMP of each element is always now
  • VTIMEZONE entries appear in random order
  • ORGANIZER CN entries randomly change between full name and plus.google.com user ID
  • ATTENDEE entries randomly change between having a CN or not having it
  • TRIGGER entries change spontaneously
  • CREATED entries change spontaneously

This causes akonadi to download and reprocess the entire ical feed at every single poll, and I can't blame akonadi for doing it. In fact, Google is saying that there is a feed with several years worth of daily appointments that all keep being changed all the time.

The work-around

As a work-around, I have configured the akonadi source to point at a local file on disk, and I have written a script to update the file only if the .ics feed has actually changed.

Have a look at the script: I consider it far from trivial, since it needs to do a partial parsing of the .ics feed to throw away all the nondeterminism that Google pollutes it with.

The setup

The script needs to be run periodically, and I used it as an opportunity to try systemd user timers:

    $ cat ~/.config/systemd/user/update-ical-feeds.timer
    [Unit]
    Description=Updates ical feeds every hour
    # Only run when on AC power
    ConditionACPower=yes

    [Timer]
    # Run every hour
    OnActiveSec=1h
    # Run a minute after boot
    OnBootSec=1m
    Unit=update-ical-feeds.service

    $ cat ~/.config/systemd/user/update-ical-feeds.service
    [Unit]
    Description=Update ICal feeds

    [Service]
    # Use oneshot to prevent two updates being run in case the previous one
    # runs for more time than the timer interval
    Type=oneshot
    ExecStart=/home/enrico/tmp/calendars/update

    $ systemctl --user start update-ical-feeds.timer
    $ systemctl --user list-timers
    NEXT                         LEFT       LAST                         PASSED UNIT                    ACTIVATES
    Wed 2015-03-25 22:19:54 CET  59min left Wed 2015-03-25 21:19:54 CET  2s ago update-ical-feeds.timer update-ical-feeds.service

    1 timers listed.
    Pass --all to see loaded but inactive timers, too.

To reload the configuration after editing: systemctl --user daemon-reload.

Further investigation

I wonder if ConditionACPower needs to be in the .timer or in the .service, since there is a [Unit] section is in both. Update: I have been told it can be in the .timer.

I also wonder if there is a way to have the timer trigger only when online. There is a network-online.target and I do not know if it is applicable. I also do not know how to ask systemd if all the preconditions are currently met for a .service/.timer to run.

Finally, I especially wonder if it is worth hoping that Google will ever make their .ics feeds play nicely with calendar clients.

Posted Wed Mar 25 21:50:21 2015 Tags:

Screen-dependent window geometry

I have an external monitor for my laptop in my work desk at home, and when I work I keep a few windows like IRC on my laptop screen, and everything else on the external monitor. Then maybe I transfer on the sofa to watch a movie or in the kitchen to cook, and I unplug from the external monitor to bring the laptop with me. Then maybe I go back to the external monitor to resume working.

The result of this (with openbox) is that when I disconnect the external monitor all the windows on my external monitor get moved to the right edge of the laptop monitor, and when I reconnect the external monitor I need to rearrange them all again.

I would like to implement something that does the following:

  1. it keeps a dictionary mapping screen geometry to window geometries
  2. every time a window geometry and virtual desktop number changes, it gets recorded in the hash for the current screen geometry
  3. every time the screen geometry changes, for each window, if there was a saved window geometry + wirtual desktop number for it for the new screen geometry, it gets restored.

Questions:

  1. Is anything like this already implemented? Where?
  2. If not, what would be a convenient way to implement it myself, ideally in a wmctrl-like way that does not depend on a specific WM?

Note: I am not interested in switching to a different WM unless it is openbox with this feature implemented in it.

Posted Mon Mar 16 21:29:36 2015 Tags:

Reuse passwords in /etc/crypttab

Today's scenario was a laptop with an SSD and a spinning disk, and the goal was to deploy a Debian system on it so that as many things as possible are encrypted.

My preferred option for it is to setup one big LUKS partition in each disk, and put a LVM2 Physical Volume inside each partition. At boot, the two LUKS partition are opened, their contents are assembled into a Volume Group, and I can have everything I want inside.

This has advantages:

  • if any of the disks breaks, the other can still be unlocked, and it should still be possible to access the LVs inside it
  • once boot has happened, any layout of LVs can be used with no further worries about encryption
  • I can use pvmove to move partitions at will between SSD and spinning disks, which means I can at anytime renegotiate the tradeoffs between speed and disk space.

However, by default this causes cryptsetup to ask for the password once for each LUKS partition, even if the passwords are the same.

Searching for ways to mitigate this gave me unsatisfactory results, like:

  • decrypt the first disk, and use a file inside it as the keyfile to decrypt the second one. But in this case if the first disk breaks, I also lose the data in the second disk.
  • reuse the LUKS session key for the first disk in the second one. Same problem as before.
  • put a detached LUKS header in /boot and use it for both disks, then make regular backups of /boot. It is an interesting option that I have not tried.

The solution that I found was something that did not show up in any of my search results, so I'm documenting it here:

    # <target name> <source device>   <key file>   <options>
    ssd             /dev/sda2         main         luks,initramfs,discard,keyscript=decrypt_keyctl
    spin            /dev/sdb1         main         luks,initramfs,keyscript=decrypt_keyctl

This caches each password for 60 seconds, so that it can be reused to unlock other devices that use it. The documentation can be found at the beginning of /lib/cryptsetup/scripts/decrypt_keyctl, beware of the leopard™.

main is an arbitrary tag used to specify which devices use the same password.

This is also useful to work easily with multiple LUKS-on-LV setups:

    # <target name> <source device>          <key file>  <options>
    home            /dev/mapper/myvg-chome   main        luks,discard,keyscript=decrypt_keyctl
    backup          /dev/mapper/myvg-cbackup main        luks,discard,keyscript=decrypt_keyctl
    swap            /dev/mapper/myvg-cswap   main        swap,discard,keyscript=decrypt_keyctl
Posted Thu Mar 12 22:45:57 2015 Tags:

Another day in the life of a poor developer

try:
    # After Python 3.3
    from collections.abc import Iterable
except ImportError:
    # This has changed in Python 3.3 (why, oh why?), reinforcing the idea that
    # the best Python version ever is still 2.7, simply because upstream has
    # promised that they won't touch it (and break it) for at least 5 more
    # years.
    from collections import Iterable

import shlex
if hasattr(shlex, "quote"):
    # New in version 3.3.
    shell_quote = shlex.quote
else:
    # Available since python 1.6 but deprecated since version 2.7: Prior to Python
    # 2.7, this function was not publicly documented. It is finally exposed
    # publicly in Python 3.3 as the quote function in the shlex module.
    #
    # Except everyone was using it, because it was the only way provided by the
    # python standard library to make a string safe for shell use
    #
    # See http://stackoverflow.com/questions/35817/how-to-escape-os-system-calls-in-python
    import pipes
    shell_quote = pipes.quote

import shutil
if hasattr(shutil, "which"):
    # New in version 3.3.
    shell_which = shutil.which
else:
    # Available since python 1.6:
    # http://stackoverflow.com/questions/377017/test-if-executable-exists-in-python
    from distutils.spawn import find_executable
    shell_which = find_executable
Posted Fri Feb 27 12:02:33 2015 Tags:

Akonadi client example

After many failed attemps I have managed to build a C++ akonadi client. It has felt like one of the most frustrating programming experiences of my whole life, so I'm sharing the results hoping to spare others from all the suffering.

First thing first, akonadi client libraries are not in libakonadi-dev but in kdepimlibs5-dev, even if kdepimlibs5-dev does not show in apt-cache search akonadi.

Then, kdepimlibs is built with Qt4. If your application uses Qt5 (mine was) you need to port it back to Qt4 if you want to talk to Akonadi.

Then, kdepimlibs does not seem to support qmake and does not ship pkg-config .pc files, and if you want to use kdepimlibs your build system needs to be cmake. I ported by code from qmake to cmake, and now qtcreator wants me to run cmake by hand every time I change the CMakeLists.txt file, and it stopped allowing to add, rename or delete sources.

Finally, most of the code / build system snippets found on the internet seem flawed in a way or another, because the build toolchain of Qt/KDE applications has undergone several redesignins during time, and the network is littered with examples from different eras. The way to obtain template code to start a Qt/KDE project is to use kapptemplate. I have found no getting started tutorial on the internet that said "do not just copy the snippets from here, run kapptemplate instead so you get them up to date".

kapptemplate supports building an "Akonadi Resource" and an "Akonadi Serializer", but it does not support generating template code for an akonadi client. That left me with the feeling that I was dealing with some software that wants to be developed but does not want to be used.

Anyway, now an example of how to interrogate Akonadi exists as is on the internet. I hope that all the tears of blood that I cried this morning have not been cried in vain.

Posted Mon Feb 23 15:44:01 2015 Tags:

The wonders of missing documentation

Update: I have managed to build an example Akonadi client application.

I'm new here, I want to make a simple C++ GUI app that pops up a QCalendarWidget which my local Akonadi has appointments.

I open qtcreator, create a new app, hack away for a while, then of course I get undefined references for all Akonadi symbols, since I didn't tell the build system that I'm building with akonadi. Ok.

How do I tell the build system that I'm building with akonadi? After 20 minutes of frantic looking around the internet, I still have no idea.

There is a package called libakonadi-dev which does not seem to have anything to do with this. That page mentions everything about making applications with Akonadi except how to build them.

There is a package called kdepimlibs5-dev which looks promising: it has no .a files but it does have haders and cmake files. However, qtcreator is only integrated with qmake, and I would really like the handholding of an IDE at this stage.

I put something together naively doing just what looked right, and I managed to get an application that segfaults before main() is even called:

/*
 * Copyright © 2015 Enrico Zini <enrico@enricozini.org>
 *
 * This work is free. You can redistribute it and/or modify it under the
 * terms of the Do What The Fuck You Want To Public License, Version 2,
 * as published by Sam Hocevar. See the COPYING file for more details.
 */
#include <QDebug>

int main(int argc, char *argv[])
{
    qDebug() << "BEGIN";
    return 0;
}
QT       += core gui widgets
CONFIG += c++11

TARGET = wtf
TEMPLATE = app

LIBS += -lkdecore -lakonadi-kde

SOURCES += wtf.cpp

I didn't achieve what I wanted, but I feel like I achieved something magical and beautiful after all.

I shall now perform some haruspicy on those oscure cmake files to see if I can figure something out. But seriously, people?

Posted Mon Feb 23 11:36:18 2015 Tags:

Setting up Akonadi

Now that I have a CalDAV server that syncs with my phone I would like to use it from my desktop.

It looks like akonadi is able to sync with CalDAV servers, so I'm giving it a try.

First thing first is to give a meaning to the arbitrary name of this thing. Wikipedia says it is the oracle goddess of justice in Ghana. That still does not hint at all at personal information servers, but seems quite nice. Ok. I gave up with software having purpose-related names ages ago.

# apt-get install akonadi-server akonadi-backend-postgresql

Akonadi wants a SQL database as a backend. By default it uses MySQL, but I had enough of MySQL ages ago.

I tried SQLite but the performance with it is terrible. Terrible as in, it takes 2 minutes between adding a calendar entry and having it show up in the calendar. I'm fascinated by how Akonadi manages to use SQLite so badly, but since I currently just want to get a job done, next in line is PostgreSQL:

# su - postgres
$ createuser enrico
$ psql postgres
postgres=# alter user enrico createdb;

Then as enrico:

$ createdb akonadi-enrico
$ cat <<EOT > ~/.config/akonadi/akonadiserverrc
[%General]
Driver=QPSQL

[QPSQL]
Name=akonadi-enrico
StartServer=false
Host=
Options=
ServerPath=
InitDbPath=

I can now use kontact to connect Akonadi to my CalDAV server and it works nicely, both with calendar and with addressbook entries.

KDE has at least two clients for Akonadi: Kontact, which is a kitchen sink application similar to Evolution, and KOrganizer, which is just the calendar and scheduling component of Kontact.

Both work decently, and KOrganizer has a pretty decent startup time. I now have a usable desktop PIM application that is synced with my phone. W00T!

Next step is to port my swift little calendar display tool to use Akonadi as a back-end.

Posted Tue Feb 17 15:34:55 2015 Tags:

seat-inspect

Four months ago I wrote this somewhere:

Seeing a DD saying "this new dbus stuff scares me" would make most debian users scared. Seeing a DD who has an idea of what is going on, and who can explain it, would be an interesting and exciting experience.

So, let's be exemplary, competent and patient. Or at least, competent. Some may like or not like the changes, but do we all understand what is going on? Will we all be able to support our friends and customers running jessie?

I confess that although I understand the need for it, I don't feel competent enough to support systemd-based machines right now.

So, are we maybe in need of help, cheat sheets, arsenals of one-liners, diagnostic tools?

Maybe a round of posts on -planet like "one debian package a day" but with new features that jessie will have, and how to understand them and take advantage of them?

That was four months ago. In the meantime, I did some work, and it got better for me.

Yesterday, however, I've seen an experienced Linux person frustrated because the shutdown function of the desktop was doing nothing whatsoever. Today I found John Goerzen's post on planet.

I felt like some more diagnostic tools were needed, so I spent the day making seat-inspect.

seat-inspect tries to make the status of the login/seat system visible, to help with understanding and troubleshooting.

The intent of running the code is to have an overview of the system status, both to see what the new facilities are about, and to figure out if there is something out of place.

The intent of reading the code is to have an idea of how to use these facilities: the code has been written to be straightforward and is annotated with relevant bits from the logind API documentation.

seat-inspect is not a finished tool, but a starting point. I put it on github hoping that people will fork it and add their own extra sanity checks and warnings, so that it can grow into a standard thing to run if a system acts weird.

As it is now, it should be able to issue warnings if some bits are missing for network-manager or shutdown functions to work correctly. I haven't really tested that, though, because I don't have a system at hand where they are currently not working fine.

Another nice thing of it is that when running seat-inspect -v you get a dump of what logind/consolekit think about your system. I found it an interesting way to explore the new functionalities that we recently grew. The same can be done, and in more details, with loginctl calls, but I lacked a summary.

After writing this I feel a bit more competent, probably enough to sit at somebody's computer and poke into loginctl bits. I highly recommend the experience.

Posted Tue Feb 10 18:06:43 2015 Tags:

Playing with python, terminfo and command output

I am experimenting with showing progress on the terminal for a subcommand that is being run, showing what is happening without scrolling away the output of the main program, and I came out with this little toy. It shows the last X lines of a subcommand output, then gets rid of everything after the command has ended.

Usability-wise, it feels like a tease to me: it looks like I'm being shown all sorts of information then they are taken away from me before I managed to make sense of them. However, I find it cute enough to share:

#!/usr/bin/env python3
#coding: utf-8
# Copyright 2015 Enrico Zini <enrico@enricozini.org>.  Licensed under the terms
# of the GNU General Public License, version 2 or any later version.

import argparse
import fcntl
import select
import curses
import contextlib
import subprocess
import os
import sys
import collections
import shlex
import shutil
import logging

def stream_output(proc):
    """
    Take a subprocess.Popen object and generate its output, line by line,
    annotated with "stdout" or "stderr". At process termination it generates
    one last element: ("result", return_code) with the return code of the
    process.
    """
    fds = [proc.stdout, proc.stderr]
    bufs = [b"", b""]
    types = ["stdout", "stderr"]
    # Set both pipes as non-blocking
    for fd in fds:
        fcntl.fcntl(fd, fcntl.F_SETFL, os.O_NONBLOCK)
    # Multiplex stdout and stderr with different prefixes
    while len(fds) > 0:
        s = select.select(fds, (), ())
        for fd in s[0]:
            idx = fds.index(fd)
            buf = fd.read()
            if len(buf) == 0:
                fds.pop(idx)
                if len(bufs[idx]) != 0:
                    yield types[idx], bufs.pop(idx)
                types.pop(idx)
            else:
                bufs[idx] += buf
                lines = bufs[idx].split(b"\n")
                bufs[idx] = lines.pop()
                for l in lines:
                    yield types[idx], l
    res = proc.wait()
    yield "result", res

@contextlib.contextmanager
def miniscreen(has_fancyterm, name, maxlines=3, silent=False):
    """
    Show the output of a process scrolling in a portion of the screen.

    has_fancyterm: true if the terminal supports fancy features; if false, just
    write lines to standard output

    name: name of the process being run, to use as a header

    maxlines: maximum height of the miniscreen

    silent: do nothing whatsoever, used to disable this without needing to
            change the code structure

    Usage:
        with miniscreen(True, "my process", 5) as print_line:
            for i in range(10):
                print_line(("stdout", "stderr")[i % 2], "Line #{}".format(i))
    """
    if not silent and has_fancyterm:
        # Discover all the terminal control sequences that we need
        output_normal = str(curses.tigetstr("sgr0"), "ascii")
        output_up = str(curses.tigetstr("cuu1"), "ascii")
        output_clreol = str(curses.tigetstr("el"), "ascii")
        cols, lines = shutil.get_terminal_size()
        output_width = cols

        fg_color = (curses.tigetstr("setaf") or
                    curses.tigetstr("setf") or "")
        sys.stdout.write(str(curses.tparm(fg_color, 6), "ascii"))

        output_lines = collections.deque(maxlen=maxlines)

        def print_lines():
            """
            Print the lines in our buffer, then move back to the beginning
            """
            sys.stdout.write("{} progress:".format(name))
            sys.stdout.write(output_clreol)
            for msg in output_lines:
                sys.stdout.write("\n")
                sys.stdout.write(msg)
                sys.stdout.write(output_clreol)
            sys.stdout.write(output_up * len(output_lines))
            sys.stdout.write("\r")

        try:
            print_lines()

            def _progress_line(type, line):
                """
                Print a new line to the miniscreen
                """
                # Add the new line to our output buffer
                msg = "{} {}".format("." if type == "stdout" else "!", line)
                if len(msg) > output_width - 4:
                    msg = msg[:output_width - 4] + "..."
                output_lines.append(msg)
                # Update the miniscreen
                print_lines()

            yield _progress_line

            # Clear the miniscreen by filling our ring buffer with empty lines
            # then printing them out
            for i in range(maxlines):
                output_lines.append("")
            print_lines()
        finally:
            sys.stdout.write(output_normal)
    elif not silent:
        def _progress_line(type, line):
            print("{}: {}".format(type, line))
        yield _progress_line
    else:
        def _progress_line(type, line):
            pass
        yield _progress_line

def run_command_fancy(name, cmd, env=None, logfd=None, fancy=True, debug=False):
    quoted_cmd = " ".join(shlex.quote(x) for x in cmd)
    log.info("%s running command %s", name, quoted_cmd)
    if logfd: print("runcmd:", quoted_cmd, file=logfd)

    # Run the script itself on an empty environment, so that what was
    # documented is exactly what was run
    proc = subprocess.Popen(cmd, stdout=subprocess.PIPE, stderr=subprocess.PIPE, env=env)

    with miniscreen(fancy, name, silent=debug) as progress:
        stderr = []
        for type, val in stream_output(proc):
            if type == "stdout":
                val = val.decode("utf-8")
                if logfd: print("stdout:", val, file=logfd)
                log.debug("%s stdout: %s", name, val)
                progress(type, val)
            elif type == "stderr":
                val = val.decode("utf-8")
                if logfd: print("stderr:", val, file=logfd)
                stderr.append(val)
                log.debug("%s stderr: %s", name, val)
                progress(type, val)
            elif type == "result":
                if logfd: print("retval:", val, file=logfd)
                log.debug("%s retval: %d", name, val)
                retval = val

    if retval != 0:
        lastlines = min(len(stderr), 5)
        log.error("%s exited with code %s", name, retval)
        log.error("Last %d lines of standard error:", lastlines)
        for line in stderr[-lastlines:]:
            log.error("%s: %s", name, line)

    return retval


parser = argparse.ArgumentParser(description="run a command showing only a portion of its output")
parser.add_argument("--logfile", action="store", help="specify a file where the full execution log will be written")
parser.add_argument("--debug", action="store_true", help="debugging output on the terminal")
parser.add_argument("--verbose", action="store_true", help="verbose output on the terminal")
parser.add_argument("command", nargs="*", help="command to run")
args = parser.parse_args()

if args.debug:
    loglevel = logging.DEBUG
elif args.verbose:
    loglevel = logging.INFO
else:
    loglevel = logging.WARN
logging.basicConfig(level=loglevel, stream=sys.stderr)
log = logging.getLogger()

fancy = False
if not args.debug and sys.stdout.isatty():
    curses.setupterm()
    if curses.tigetnum("colors") > 0:
        fancy = True

if args.logfile:
    logfd = open("output.log", "wt")
else:
    logfd = None

retval = run_command_fancy("miniscreen example", args.command, logfd=logfd)

sys.exit(retval)
Posted Wed Jan 21 11:13:31 2015 Tags:

Non importa che mi dai del voi

Dai, non importa che mi dai del voi

In che senso?

Eh, mi dici sempre "voi informatici", "voi tecnici", "voi..."

Posted Fri Dec 19 15:55:20 2014 Tags:
Posted Sat Jun 6 00:57:39 2009

Localising free software for Taiwanese Aboriginal cultures.

People who participated so far:

Character list for the Paiwan language

We mapped the available glyphs and accents for the Paiwan language.

The letters in alphabetical order:

a b c d e f h i j k l m n p q r s t u v w y z ḏ nġ ḻ ṟ ṯ 

No uppercase.

Update: this character list has been improved and the good version is found in the Debian wiki.

All the characters are in Unicode except nġ, which already needs to be requested for the Amis script.

We need to design an input method to enter the underlined letters and the nġ.

Update: there is now a wiki page on the Debian wiki.

Posted Sat Jun 6 00:57:39 2009 Tags:

Creating a new locale

I'm currently in Cilamitay, in the east of Taiwan. There is a little meeting of Taiwanese Free Software people and people from the Amis, Taroko and Puyuma tribes, with the idea of starting localisation efforts for some aboriginal languages.

These are some of the issues we are going to discuss:

Language code

A new ISO standard (639-3) will hopefully be formalised in January that will include the language codes for the Taiwanese aboriginal tribes. We'll have to work some temporary solution, but there's good hope that it won't have to be temporary for long.

List of characters

Because of Christian missionary influence, both Amis and Taroko use a roman alphabet, with accents. We need to work out the complete list of character and accent combination, see if everything is in Unicode, see how they sort.

We then need to find a comfortable way to input them using the keyboards normally available here (English US layout): compose key? Dead keys? How about on Windows?

Womble2 on IRC tells me that on Windows one can works with MSKLC.

Technical terms and country list

We need to work out how to map terms that do not exist in the language.

Technical terms are usually borrowed from Japanese.

Names for all the countries in the world probably do not exist.

Translation interface

We need to find an easy to use interface to input the translations.

There is Rosetta.

There is Pootle. (Thanks to Christian Perrier for pointing me at it)

There is Webpot.

Update: there is now a wiki page on the Debian wiki.

Posted Sat Jun 6 00:57:39 2009 Tags:

Happy new year

A year ago we got in touch with various Taiwanese aboriginal tribes to try to start localisation efforts.

Thanks to the research the Taroko people did during 2007 and the prototype work of tonight, the Taroko people in Taiwan can see the computer calendar of the new year in their own language:

trv_TZW Gnome calendar

Posted Sat Jun 6 00:57:39 2009 Tags:

Amis and Paiwan input method and character set

Arne Götje (高盛華) created:

The scripts, especially Amis, make heavy use of Unicode combination characters. They should display well at least with the Dejavu Sans font in many applications.

Try it out: if it displays correctly, you should see:

  • accented letters instead of letters next to accents.
  • i with both the dot and the accent.

Update: there is now a wiki page on the Debian wiki.

Posted Sat Jun 6 00:57:39 2009 Tags:

Character list for the Amis language

We mapped the available glyphs and accents for the Amis language.

The letters in alphabetical order:

    a c d f ng h i k l m n o p r s t u w y

Everyone of them can get an acute or circumflex accent on top. ng can get a dot on top of the g.

The accents are literally on top: i would get the dot PLUS the accent on top.

Not all accented characters directly exist in Unicode; however Unicode developed various kinds of combination features to take care of these cases.

Then we need an input method that would insert ng instead of g and allow to type all the accent combinations.

Here is the full character set:

    a     á    â
    c     ć    ĉ
    d     d́    d̂
    f     f́    f̂
    ng    nǵ   nĝ  nġ
    h     h́    ĥ
    i     i̇́    i̇̂
    k     ḱ    k̂
    l     ĺ    l̂
    m     ḿ    m̂
    n     ń    n̂
    o     ó    ô
    p     ṕ    p̂
    r     ŕ    r̂
    s     ś    ŝ
    t     t́    t̂
    u     ú    û
    w     ẃ    ŵ
    y     ý    ŷ

Update: this character list has been improved and the good version is found in the Debian wiki.

The list is not displayed correctly with many fonts or rendering engines. Arne made a test page that explicitly sets a font that works.

The accents are not taken into account when sorting.

Uppercase letters are not used.

Note: the page has been updated to reflect further input from Unicode and Amis people.

Update: there is now a wiki page on the Debian wiki.

Posted Sat Jun 6 00:57:39 2009 Tags:
Posted Sat Jun 6 00:57:39 2009

Cazzeggio.

Non importa che mi dai del voi

Dai, non importa che mi dai del voi

In che senso?

Eh, mi dici sempre "voi informatici", "voi tecnici", "voi..."

Posted Fri Dec 19 15:55:20 2014 Tags:

Spelling a chilometri zero

Lo spelling internazionale è troppo globalizzato, e volete recuperare un attimo la dimensione del posto dove siete nati e cresciuti?

Da oggi c'è questo script che fa per voi: gli dite dove abitate, e lui vi crea lo spelling a chilometri zero.

$ git clone git@gitorious.org:trespolo/osmspell.git
$ cd osmspell
$ ./osmspell "San Giorgio di Piano"
1: San Giorgio di Piano, BO, EMR, Italia
2: San Giorgio di Piano, Via Codronchi, San Giorgio di Piano, BO, EMR, Italia
3: San Giorgio Di Piano, Via Libertà, San Giorgio di Piano, BO, EMR, Italia
Choose one: 1
Center: 44.6465332, 11.3790398
A Argelato, Altedo
B Bentivoglio, Bologna, Boschi
C Cinquanta, Castagnolo Minore, Castel Maggiore, Cento
D Dosso
E Eremo di Tizzano
F Funo di Argelato, Finale Emilia, Ferrara, Fiesso
G Gherghenzano, Galliera, Gesso
I Il Cucco, Irnerio, Idice
L Località Fortuna, Lovoleto, Lippo
M Malacappa, Massumatico, Minerbio, Marano
N Navile
O Osteriola, Ozzano dell'Emilia, Oca
P Piombino, Padulle, Poggio Renatico, Piave
Q Quarto Inferiore, Quattrina
R Rubizzano, Renazzo, Riale
S San Giorgio di Piano, Saletto
T Torre Verde, Tintoria, Tombe
U Uccellino
V Venezzano Mascarino, Vigarano Mainarda, Veduro
X XII Morelli
Z Zenerigolo, Zola Predosa

I dati vengono da OSM, e lo script è un ottimo esempio di come usarne la API di geolocazione (veloci) e la API di query geografica (lenta).

Posted Sat Jan 4 00:38:16 2014 Tags:

Poesia: "Lavatrice"

Pensavo fosse pail,

invece ora è feltro.

Posted Tue Dec 3 22:32:23 2013 Tags:

Shops

Christmas songs should only ever be played on Christmas day.

In church.

At midnight.

Unless I happen to be there.

Posted Mon Dec 2 14:07:58 2013 Tags:

Airports

Photo of a commercial in London City airport saying 'In the lap of luxury - Want to reach a captive audience with dwell time? Why advertise anywhere else? - London City Airport Media Sales'

In the airport, we are not travellers. We are a captive audience with dwell time.

In other words, suckers stuck in a room where the only pastime provided is spending money and staring at advertisements selling advertisement space in rooms full of suckers like them.

Posted Fri Nov 22 18:58:00 2013 Tags:

Explanation of umarell

Umarell /uma'rɛl/ (oo-mah-rell), n; pl. Umarells. People in a community who offer all sorts of comments to those who are trying to get some work done, but who are not doing any work themselves.

Etymology and further details

Umarell is a word that entered Italian slang in Bologna and is spreading to nearby towns, occasionally even across Italy. It comes from the Bolognese for "cute/odd little man".

"Umarells" are those people, usually retired men, who spend time watching construction works, often holding their hands behind their back, occasionally commenting on what is going on, sometimes trying to tell the workers what to do.

It's easy to find examples on the internet; the word was popularised by a blog collecting photos, which has even been published into a book.

With some Italian Debian friends, we realised that umarell is the perfect word to describe those people in a community, who offer all sorts of comments to those who are trying to get some work done, but who are not doing any work themselves.

I think that it is a word that fits perfectly, and since I'm likely going to use it blissfully anywhere, here is a page that temporarily explains what it means until the Oxford English Dictionary picks it up.

Posted Fri Sep 20 13:27:07 2013 Tags:

Yet another Ubuntu anecdote

Some posts on planet made me remember of a little Canonical-related story of mine.

Many years ago I shortly contracted for Canonical. It was interesting and fun.

At the time I didn't have any experience of being temporarily hired by a foreign company, so I rang my labour union to get an appointment, to make sure with them that everything was allright.

The phone call went more or less like this:

Me:

Hello. I have received this contract for temporary employment by a foreign company and I wondered if I could book an appointment to come show it to you to see if it's all ok.

Their answer rather cut me short:

Hi. Be careful! People get temporary employment from obscure companies with the headquarters, like, in the Isle of Man, they do the job, the company disappears and they never get paid. There's bad stuff out there!

I looked at the contract, the heading said something like "Canonical ltd, Douglas, Isle of Man".

I was certain that the union people would have never understood what was going on. I politely thanked them for their time and hung up. However, to this day I still regret that I didn't insist:

Uh, yes, the company is indeed in the Isle of Man. But what if I told you that it's owned by an astronaut?

I just signed the contract and had a good time.

Posted Sat Jan 15 10:35:36 2011 Tags:

Mailman defaults

Monopoly Chance: It's the first of the month. / You're flooded with mailman junk / Skip a turn.

Posted Fri Oct 1 12:03:53 2010 Tags:

My rule to see if a framework is worth of attention

I came up with a little rule:

In order to be worth of any attention, a framework must be stable enough that I can charge money to train people to use it.

This probably applies to other kinds of software stacks, libraries, development environments and, well, to most software applications.

In the context of python web frameworks, this means that:

  • If it changes API all the time it is not worth of attention, because my customers won't get value for their money, as they'd continuously need retraining and rewriting their software.
  • If I see lots of DeprecationWarnings it is not worth of attention, because my customers will see them and blame me for teaching them deprecated stuff.
  • If fixes for bugs affecting the stable version are only distributed "in a recent git" or "in the next development version", and they are not backported into a new bugfix-only stable release, then it is not worth of attention, because:
    • my customers' business is to develop their own products based on the framework.
    • My customers' business is not to be maintaning in-house stable updates of the framework. Although if the framework's community is nice enough they might end up giving a hand.
  • If it requires virtualenv or can only be obtained through easy_install it is not worth of attention, because:
    • my customers are not interested in maintaning custom deployment environments over time.
    • My customers are not interested in tracking each and every single library's upstream development to keep their production system free of bugs.
    • My customers are used to getting software through a proper distribution which also takes care of security updates.
    • I am paid to teach them how to use a framework, not a custom python-only package management system.
    • In my experience, if distributions have trouble keeping packages up to date, upstream is doing something fundamentally wrong.

In light of this rule, I regret to notice that I see very few python web frameworks worth of any attention.

Posted Wed Aug 4 15:32:24 2010 Tags:

On python stable APIs

There is a theory which states that if ever anyone discovers exactly what the Universe is for and why it is here, it will instantly disappear and be replaced by something even more bizarre and inexplicable.

There is another theory which states that this has already happened.

In Debian testing:

/usr/lib/python2.6/dist-packages/sqlalchemy/types.py:547: SADeprecationWarning: The Binary type has been renamed to LargeBinary.

In Debian Lenny:

ImportError: cannot import name LargeBinary

I was starting to think that SQLAlchemy wasn't too bad, since I've been using it for 6 months and I haven't seen its API change yet.

But there it is, a beautiful reminder that SQLAlchemy, too, is part of the marvelously autistic Python ecosystem.

Posted Mon Jul 19 16:14:25 2010 Tags:
Posted Sat Jun 6 00:57:39 2009