Bug#552528: Caused by character set mismatch

[Matt Kraai]

On Mon, Nov 09, 2009 at 11:55:57PM +0100, Josselin Mouette wrote:
> Le samedi 07 novembre 2009 à 13:32 -0800, Matt Kraai a écrit : 
> > On my system, alacarte has a segmentation fault if I reconfigure the
> > locales package to generate the it_IT at euro locale and set LANG to
> > "it_IT at euro".
> > 
> > While it's parsing the UI definition, it translates "Restore the
> > default menu layout" to "Ripristina la struttura predefinita del
> > men�", which uses the ISO/IEC 8859-15 character set.  It subsequently
> > calls g_markup_escape_text on this string, which expects it to use
> > UTF-8.  This causes it to miss the end of the string and, eventually,
> > to try to access an inaccessible page, causing the segmentation fault.
> > 
> > I don't know where or how this should be fixed.
> 
> There is probably a missing bind_textdomain_codeset call, or whatever
> the equivalent is in Python.

Python's gettext module provides a bind_textdomain_codeset call, but
adding

 gettext.bind_textdomain_codeset(config.GETTEXT_PACKAGE,'UTF-8')

and prefixing GETTEXT_PACKAGE on the previous two lines with "config."
doesn't eliminate the segmentation fault.

> Anyway, we don’t support non-UTF8 locales anymore, so that’s not a very
> important bug.

Would it be OK to downgrade its severity?  If so, to what?

-- 
Matt                                            http://ftbfs.org/kraai