py-matplotlib gdk backend broken

classic Classic list List threaded Threaded
4 messages Options
Reply | Threaded
Open this post in threaded view
|

py-matplotlib gdk backend broken

Edd Barrett-3
Hi,

I was using matplotlib at the end of last week to build some graphs for
a presentation.

I upgraded my snapshot over the weekend, and now I'm getting a crashes.
E.g. the xkcd example graph [1] worked fine on Thursday, but now:

---8<---
$ python2.7 xkcd.py            
Traceback (most recent call last):
  File "xkcd.py", line 1, in <module>
    from matplotlib import pyplot as plt
  File "/usr/local/lib/python2.7/site-packages/matplotlib/pyplot.py", line 109, in <module>
    _backend_mod, new_figure_manager, draw_if_interactive, _show = pylab_setup()
  File "/usr/local/lib/python2.7/site-packages/matplotlib/backends/__init__.py", line 32, in pylab_setup
    globals(),locals(),[backend_name],0)
  File "/usr/local/lib/python2.7/site-packages/matplotlib/backends/backend_gtkagg.py", line 14, in <module>
    from matplotlib.backends.backend_gtk import gtk, FigureManagerGTK, FigureCanvasGTK,\
  File "/usr/local/lib/python2.7/site-packages/matplotlib/backends/backend_gtk.py", line 36, in <module>
    from matplotlib.backends.backend_gdk import RendererGDK, FigureCanvasGDK
  File "/usr/local/lib/python2.7/site-packages/matplotlib/backends/backend_gdk.py", line 33, in <module>
    from matplotlib.backends._backend_gdk import pixbuf_get_pixels_array
ImportError: No module named _backend_gdk
--->8---

matplotlib hasn't changed, so I guess something GTKish has changed.

This only seems to affect the gdk backend. For example, before importing
pyplot, you can add:

---8<---
import matplotlib                                                              
matplotlib.use(some-different-backend)
--->8---

And this will work. List of backends at [2].

I'm currently working around the crash using the "agg" backend to build
my graphs.

[1] http://matplotlib.org/xkcd/examples/showcase/xkcd.html
[2] http://matplotlib.org/faq/usage_faq.html#what-is-a-backend

--
Best Regards
Edd Barrett

http://www.theunixzoo.co.uk

Reply | Threaded
Open this post in threaded view
|

Re: py-matplotlib gdk backend broken

Stuart Henderson
On 2016/03/29 10:14, Edd Barrett wrote:
> Hi,
>
> I was using matplotlib at the end of last week to build some graphs for
> a presentation.
>
> I upgraded my snapshot over the weekend, and now I'm getting a crashes.
> E.g. the xkcd example graph [1] worked fine on Thursday, but now:

Reinstalling py-matplotlib fixes it, I've bumped REVISION to make
that happen. Would be nice if someone can figure out what's causing
it, maybe it needs an extra WANTLIB on something to force updates.

Reply | Threaded
Open this post in threaded view
|

Re: py-matplotlib gdk backend broken

Laurence Tratt
On Tue, Mar 29, 2016 at 10:45:59AM +0100, Stuart Henderson wrote:

Hello Stuart,

> Reinstalling py-matplotlib fixes it, I've bumped REVISION to make that
> happen. Would be nice if someone can figure out what's causing it, maybe it
> needs an extra WANTLIB on something to force updates.

Late last week after a snapshot update, plus packages upgrade, redshift (at
least, redshift-gtk) didn't work. The problem there seemed to be the
reference to the py3-gobject3 package. Deleting that (which, of course, also
deletes redshift) and reinstalling both packages fixed the problem.


Laurie
--
Personal                                             http://tratt.net/laurie/
Software Development Team                                http://soft-dev.org/
   https://github.com/ltratt              http://twitter.com/laurencetratt

Reply | Threaded
Open this post in threaded view
|

Re: py-matplotlib gdk backend broken

Stuart Henderson
On 2016/03/30 16:07, Laurence Tratt wrote:

> On Tue, Mar 29, 2016 at 10:45:59AM +0100, Stuart Henderson wrote:
>
> Hello Stuart,
>
> > Reinstalling py-matplotlib fixes it, I've bumped REVISION to make that
> > happen. Would be nice if someone can figure out what's causing it, maybe it
> > needs an extra WANTLIB on something to force updates.
>
> Late last week after a snapshot update, plus packages upgrade, redshift (at
> least, redshift-gtk) didn't work. The problem there seemed to be the
> reference to the py3-gobject3 package. Deleting that (which, of course, also
> deletes redshift) and reinstalling both packages fixed the problem.

I think this may be tied in with the pthread changes, libffi, etc.
The best approach for now is probably wait for the dust to settle with
the second round of bumps which has just occurred and see if people
still have problems, then decide whether to just bump, or add some
'Extra' deps somewhere, or something else.

If people do run into problems, make sure that packages are updated
for the the base libraries (they currently *won't* be - expect them
for fast arch in a couple of days) - if still problems, drop a mail
to ports@ details what, please include output from ldd if it involves
a binary program, and output with LD_DEBUG set may be useful if it's
something like python, you'll likely be able to unbreak things by
reinstalling some package or other, but it would be useful for us
to have the information about what breaks.