[Telepathy] Accessing Folks from python

[Philip Withnall]

On Mon, 2011-12-05 at 21:05 +0100, Frederik Elwert wrote:
> Am Montag, den 05.12.2011, 19:17 +0000 schrieb Philip Withnall:
> > On Mon, 2011-12-05 at 19:15 +0100, Frederik Elwert wrote:
> > > > Le vendredi 02 décembre 2011 à 10:21 +0100, Frederik Elwert a écrit :
> > > > > Now since we would prefer using python for our application, I was
> > > > > wondering if it is possible to access Folks from python. I could manage
> > > > > to import it from gi.repository, but it then fails during
> > > > > IndividualAggregator.prepare():
> > > > > 
> > > > >         from gi.repository import Folks
> > > > >         agg = Folks.IndividualAggregator()
> > > > >         agg.prepare()
> > > > >         Traceback (most recent call last):
> > > > >           File "<stdin>", line 1, in <module>
> > > > >           File "/usr/lib/python2.7/dist-packages/gi/types.py", line 43,
> > > > >         in function
> > > > >             return info.invoke(*args, **kwargs)
> > > > >         TypeError: prepare() takes exactly 3 arguments (1 given)
> > > > > 
> > > > > I have seen that the generated c code has this function signature:
> > > > > 
> > > > >         void folks_individual_aggregator_prepare
> > > > >         (FolksIndividualAggregator* self, GAsyncReadyCallback
> > > > >         _callback_, gpointer _user_data_);
> > > > 
> > > > I don't think you can omit arguments with g-i/pygobject. The only
> > > > annotation that we can set is (allow-none) but even then you have to
> > > > pass 'None' to it.
> > > > 
> > > > agg.prepare(None, None) should work. otoh, you probably want to set a
> > > > callback to know when you can start using that aggregator...
> > > 
> > > It somehow seems to do something, but then segfaults:
> > > 
> > >         agg.prepare(None, None)
> > >         (process:5627): folks-DEBUG: Using built-in backends key file '/home/frederik/.local/share/folks/backends.ini' (override with environment variable FOLKS_BACKEND_STORE_KEY_FILE_PATH)
> > >         Speicherzugriffsfehler [segmentation fault in German]
> > > 
> > > I also get a segfault when passing a python function as callback.
> > 
> > Please provide a stacktrace (either here or as a bug against folks on
> > bugzilla.gnome.org) and we can take a look.
> 
> I attached a python test file and the gdb output. (I’m not too familiar
> with debugging, so if you need additional information, please let me
> know.)

The stacktrace wasn't too useful without libfolks debug symbols, but I
could reproduce the problem myself with the program you provided.

This looks like a bug in the way folks handles construction of
IndividualAggregators which doesn't show up when folks is used from C or
Vala, but is quite obvious when folks is used from Python.

I've opened https://bugzilla.gnome.org/show_bug.cgi?id=665692 about it,
and I'm working on a patch now. I've CCed you to the bug (I hope you
don't mind).

Philip

> > > It might be because I’m rather new to GI, but I am a bit confused by the
> > > API difference between C (prepare taking parameters) and Vala (prepare
> > > as instance method taking no parameters). First, I looked at the Vala
> > > API and assumed that Python would have a similar one.
> > 
> > The prepare() method is async (using GAsyncResult). It takes parameters
> > in C because that's the only way async methods can be implemented in C.
> > In Vala, however, the ‘async’ keyword is used instead. Since async
> > methods aren't supported in Python it uses the same parameters as the C
> > code.
> 
> Okay, I see. Thanks for the explanation.
> 
> Regards
> Frederik
> _______________________________________________
> telepathy mailing list
> telepathy at lists.freedesktop.org
> http://lists.freedesktop.org/mailman/listinfo/telepathy

-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 230 bytes
Desc: This is a digitally signed message part
URL: <http://lists.freedesktop.org/archives/telepathy/attachments/20111206/f9b4a14b/attachment.pgp>