Re: what was the state of the python bindings back in 0.8?

[Jiří Techet <techet gmail com>]

On Wed, Mar 30, 2011 at 02:06, Robert Park <rbpark exolucere ca> wrote:
> Hey guys, I just need a quick history lesson here.
>
> I'm becoming increasingly concerned that fedora 15 is going to ship
> with libchamplain 0.8 instead of 0.10, which means that I'll have to
> backport my app to the old 0.8 API if I want users of Fedora 15 to be
> able to use my app. So, I made an initial attempt at this today, and I
> ran into some problems.
>
> I experienced a severe bug in which portions of the map would not
> update properly, dependant on zoom level. At zoom level 0, the entire
> map was fine. At level 1, something like a narrow bar down the middle
> of the map wouldn't update, but I could pan around the map and see it
> update on either side of this bar. At level 2, the top half was frozen
> while the bottom half moved freely. At zoom level 3 and above, no
> parts of the screen would update at all. I say 'frozen', but really
> the map could be forced to redraw by resizing the window, and it would
> show updated locations of where I'd navigated to, so it was clear that
> the map was responding to input, just not drawing itself correctly.
>
> I then went to an interactive python prompt and did the simplest
> reproducer (a single window containing only the GtkChamplain.Embed()
> widget), and it behaved the same way.
>
> Now, I should mention at this point, that I started developing my app
> against libchamplain 0.4 and static bindings, and then quickly
> migrated to 0.6 and introspected bindings, and later migrated to
> jhbuild and introspected bindings, so I kinda skipped the 0.8 series.
> Was this a known bug in the 0.8 line? Is this possibly a result of
> incorrect introspection annotations?
>
> Should I bother filing a bug about this? Or should I just write off
> Fedora 15 and tell my potential users that they'll have to wait for
> Fedora 16 in order to have the necessary dependencies to run my app?

Hi Robert,

what you are experiencing is just our good old friend:

https://bugzilla.gnome.org/show_bug.cgi?id=638652

I assume Fedora ships with clutter 1.6 and older version of
libchamplain 0.8.x. The bug should be fixed in libchamplain 0.8.2.
Unfortunately I have just realized that I haven't tested the release
enough and that polygons are not displayed because they are handled
differently in 0.8. I've already fixed it and will make the 0.8.3
release with the fix just after writing this email.

If distributions are going to ship libchamplain 0.8.x with clutter
1.6, they will have to ship the 0.8.3 release otherwise libchamplain
just won't work.

About the python support in 0.8 - it should be identical to what was
in master at the beginning of January before I started making the
bigger changes in the API.

Jiri