Fwd: Abiword 2.8.1 crashiness

---------- Forwarded message ----------
From: Martin Sevior <msevior@gmail.com>
Date: Thu, Nov 19, 2009 at 10:39 AM
Subject: Re: Abiword 2.8.1 crashiness
To: "J.M. Maurer" <uwog@uwog.net>

I vote to drop gucharmap support. It's of very marginal usefulness and
definitely not worth the pain of lots of ifdef's.
Our builtin dialog works very well for fornts with large glyph
coverage. People who really need gucharmap can just copy and paste
chars from the stand alone program.

Cheers

Martin

On Thu, Nov 19, 2009 at 8:09 AM, J.M. Maurer <uwog@uwog.net> wrote:
>
> On Wed, 2009-11-18 at 17:51 +0100, Morten Juhl-Johansen Zölde-Fejér
> wrote:
>> On Mon, 9 Nov 2009 10:40:11 +0100
>> Morten Juhl-Johansen Zölde-Fejér <mjjzf@syntaktisk.dk> wrote:
>>
>> > ---- 8< ---- Martin Sevior ---- 8< ----
>> > > Hi Morten,
>> > >               I get no crashes with "insert Symbol" in my build on
>> > > Fedora 11. Can you do a --enable-debug in the configure line, then
>> > > redo your build, run it in gdb and type "bt" in gdb  when it
>> > > crashes? This will help isolate the problem. You might have a buggy
>> > > version of pango. We push pango very hard in ways many other
>> > > applications don't. Thanks!
>> > > Martin
>> > ---- 8< ---- Martin Sevior ---- 8< ----
>> >
>> > Excellent. I will give it a spin when I come from work.
>>
>> More specific:
>> It seems the program crashes when I double-click a symbol. When I click
>> once on the symbol and press Insert, there are no problems.
>
> Ah, that was the hint I needed: gucharmap has changed its API... *yet
> again*. The "activate" callback used to look like this (from 2.22.0):
>
> static void
> insert_character_in_text_to_copy (GucharmapTable  *chartable,
>                                  gunichar         wc,
>                                  GucharmapWindow *guw)
>
> The "activate" callback these days looks like (from 2.26.1):
>
> insert_character_in_text_to_copy (GucharmapChartable *chartable,
>                                  GucharmapWindow *guw)
>
> Notice the missing "gunichar" argument. Looking through the history of
> gucharmap, I see it was dropped in 2.24.0.
>
> Two ways to solve this:
>
> 1. make gucharmap version specific DEFINEs to deal with the API breakage
> 2. don't support gucharmap < 2.24, and fallback to our builtin dialog.
> 3. drop gucharmap support alltogether and fallback to our builtin
> dialog. They've broken the API before, and they will likely break it
> again in the future (API stability is not their concern last time I
> checked).
>
> Comments on which to choose? I'm leaning toward 2).
>  Marc
>
>
Received on Thu Nov 19 00:40:54 2009