Re: [Gtkpod-devel] libgpod Python: time_released field (was: gPodder + libgpod CVS = borked)

[Todd Z. <tm...@po...>]

Nicholas Piper wrote:
> Hmm... Yeah, I like this idea. Then at 0.6 we can remove the integer
> version entirely.

There may be iPhone support in libgpod then too, which will make
everyone rush out to upgrade. :)

> I'm trying to think of when you'd want the bindings, but wouldn't
> want to install pygtk. I guess an embedded device of some kind? But
> then you're unlikely to have Python itself.

There are some people that are allergic to any gtk libraries for some
reason.  I don't pretend to understand them.

As an example, the Gentoo libgpod packagers were patching this to
accommodate users who disabled gtk in their use_flags.  They recently
changed this and simply force gtk to be installed if the bindings are
requested.

If I remember, I'll try to ping the gentoo packager(s) and ask them if
they can test out these changes and let us know if they find any
remaining issues when building without gtk.  (I'd definitely like to
get more input from downstream packagers and users in general if there
are build/packaging issues that they feel the need to patch.)

>> If building without gdk-pixbuf, some of the python photo tests fail.
>> I suppose that's the way it should be (?), but I wanted to make sure
>> you and John knew.
>=20
>> FAIL: testAddCountPhotos (__main__.TestPhotoDatabase)
>> FAIL: testAddPhoto (__main__.TestPhotoDatabase)
>> FAIL: testAddPhotoToAlbum (__main__.TestPhotoDatabase)
>> FAIL: testAddRemovePhoto (__main__.TestPhotoDatabase)
>> FAIL: testRemovePhotoFromAlbum (__main__.TestPhotoDatabase)
>=20
> I guess it's fair that they don't work.

I wasn't sure.  Technically, if I disable gdk-pixbuf and then run make
test, I'd expect the tests for disabled functionality to not be run.
How hard is it to only run those tests if gdk-pixbuf is enabled?

> Could you commit this to the branch? Then we can merge that into the
> trunk and concentrate only on the time-related patch.

Done.

> Thank you for your careful attention, once again :-)

Sure.  Of course, if I'd have been more careful (and knowledgeable),
I'd have probably used:

    %typemap(out) time_t {
       $result =3D PyInt_FromLong($1);
    }

(and some corresponding %typemap(in))

instead of:

    typedef long time_t;

Then we wouldn't have this damn discrepancy between versions. :(

--=20
Todd        OpenPGP -> KeyID: 0xBEAF0CE3 | URL: www.pobox.com/~tmz/pgp
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
There are two kinds of pedestrians -- the quick and the dead.