[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Re: cdebconf_0.196_i386.changes ACCEPTED into unstable



-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA256

Hi everyone,

On 16/11/15 22:01, Cyril Brulebois wrote:
> Hi,
> 
> Samuel Thibault <sthibault@debian.org> (2015-11-16):
>> Cyril Brulebois, on Mon 16 Nov 2015 22:39:26 +0100, wrote:
>>> I'm sorry but the implementation doesn't look reasonable at
>>> first glance. We already have gtk-set-font, which handles font
>>> name and font size selection, in rootskel-gtk. Couldn't we
>>> adjust that instead of implementing a(nother) sed call within
>>> cdebconf-gtk?
>> 
>> Christian uploaded before I got ideas about doing it in a better
>> way :)
> 
> That happens. ;)
> 
>> I've been thinking about it, and perhaps we can indeed make 
>> cdebconf instead increase a value in some /etc/gtk-2 file, which 
>> gtk-set-font would add to its default "9" value. We'd then have
>> to make cdebconf_gtk_di_run_dialog run
>> gtk_rc_reparse_all_for_settings().
> 
> Something along these lines would probably work, yes.
> 
>>> Also, picking a random non-default language like Chinese or 
>>> Dzongkha, then increasing the font leads to G_IS_OBJECT
>>> assertion failure and localechooser fails with error 1 (going
>>> back to the language selection).
>> 
>> Ah, I had tested that, but didn't realize it was still posing
>> problem. This seems very odd, and it seems to me like some bug
>> deeper somewhere.  I initially just modified gtkrc and called 
>> gtk_rc_reparse_all_for_settings(), but the next choice being
>> made would get the kind of assertion failure you are talking
>> about.  I added the cdebconf_gtk_set_answer_notok() call to make
>> sure to restart the question with the new size (which is needed
>> anyway to recompute the layout).
> 
> I must confess I don't have much time to dig into this. Feel free
> to record that as a bug report for the time being; while we hadn't
> ctrl-+/- support until this upload, this might break cause
> regressions depending on user input, and cause some confusion.
> 
> Mraw, KiBi.
> 

Jumping slightly on the cdebconf bandwagon here. Having finally a
proper Internet connection, I've pushed a few changes that had been
sleeping on my laptop for a while. They include :

 * Lintian overrides for what is considered misuse of debconf templates.
 * SeLinux support, which is disabled for d-i builds
 * Rewording of some long descriptions
 * The long awaited possibility to show passwords in clear with the
GTK and newt plugins. (Sorry Christian, more translations).

Overall, the actual changes are relatively limited and local when
taking patches individually, so should be fairly safe.

If there's anything I could improve, please let me know.

Thanks,
Regis


Reply to: