gtk+2.4.0 released

Jürg Billeter j at bitron.ch
Fri Mar 19 13:37:24 PST 2004


On Fre, 2004-03-19 at 18:36, Dagmar d'Surreal wrote:
> On Fri, 2004-03-19 at 01:29, Jürg Billeter wrote:
> > On Don, 2004-03-18 at 23:14, Kelledin wrote:
> > > Presumably Gnome 2.6 will take care of most or all of this; it's 
> > > supposedly due out pretty soon.  Still, I'm considering advising 
> > > the Gnome developers to exclude those flags in "release" 
> > > packages.  That sort of thing is good for developers (the same 
> > > way "gcc -Wall -Werror" is) but causes problems for end-users 
> > > trying to compile "released" packages.  Beyond the annoyance of 
> > > having a compile break, it can also cause latent instability 
> > > even if the package appears to compile successfully.
> > 
> > This has already been done, at least most of the packages now include
> > these flags only in maintainer builds. Gnome 2.6 (as of RC1) builds of
> > course seamless without any Makefile modifications.
> 
> Actually, I spoke too soon.  After about a half-dozen recompiles...
> 
> audacity: relocation error: /usr/lib/libwx_gtk2-2.4.so: undefined
> symbol: _gtk_accel_group_attach
> 
> Traceback (most recent call last):
>   File "/usr/bin/btdownloadgui.py", line 13, in ?
>     from wxPython.wx import *
>   File "/usr/lib/python2.3/site-packages/wxPython/__init__.py", line 20,
> in ?
>     import wxc
> ImportError: /usr/lib/libwx_gtk2-2.4.so: undefined symbol:
> _gtk_accel_group_detach
> 
> The irksome thing is that I've seen commentary on the Gtk+ list (from
> some prudent Googling) to the effect that this function isn't something
> wxGTK should be messing with.
Yeah, it looks like wxGTK uses a function of gtk+ which has been marked
as private, so it can change between minor releases as applications
shouldn't use it directly.





More information about the blfs-dev mailing list