Re: gtksourceviewmm issues.

[Murray Cumming <murrayc murrayc com>]

On Tue, 2010-11-09 at 16:10 +0100, Krzesimir Nowak wrote:
> Hi,
> 
> I recently tried to turn gtksourceviewmm-3.0 into some sensible shape
> and I maybe partially succeeded, that is - all TODOs in source files are
> handled, completion stuff seems to work.
> 
> However there are several issues I'm leaving as they are for now:
> 
> ---
> 1. Search test (in tests/search directory) segfaults in line 53:
> 
> get_vbox ()->pack_start (*m_hbox) ;
> 
> To reproduce it, just run the test and do Edit->Search.

Fixed in gtkmm 3:
http://git.gnome.org/browse/gtkmm/commit/?id=fa24dac983b002c9e2371c6cc82587d3bb282ed7

> ---
> 2. Basic test (in tests/basic) aborts with failed assertion:
> 
> Gtk:ERROR:gtktextview.c:3002:gtk_text_view_finalize: assertion failed:
> (priv->buffer == NULL)
> 
> To reproduce it just run the test in hit the X in the corner of title
> bar to quit the application. In general this assertion pops when
> everything is being destroyed.

Yes. I don't know why yet. Here's a backtrace from gdb with "break
g_log":

#0  0x0012e416 in __kernel_vsyscall ()
#1  0x01738941 in raise (sig=6)
at ../nptl/sysdeps/unix/sysv/linux/raise.c:64
#2  0x0173be42 in abort () at abort.c:92
#3  0x015421e1 in g_assertion_message (domain=0xdad29a "Gtk",
file=0xdad28c "gtktextview.c", line=3002, func=0xdae9be
"gtk_text_view_finalize", 
    message=0x80c6810 "assertion failed: (priv->buffer == NULL)") at
gtestutils.c:1358
#4  0x01542244 in g_assertion_message_expr (domain=0xdad29a "Gtk",
file=0xdad28c "gtktextview.c", line=3002, func=0xdae9be
"gtk_text_view_finalize", 
    expr=0xdad9f7 "priv->buffer == NULL") at gtestutils.c:1369
#5  0x00c1df12 in gtk_text_view_finalize (object=0x80c5110) at
gtktextview.c:3002
#6  0x0082e31e in gtk_source_view_finalize (object=0x80c5110) at
gtksourceview.c:1747
#7  0x0147af93 in g_object_unref (_object=0x80c5110) at gobject.c:2713
#8  0x0046f580 in Gtk::Object::_release_c_instance (this=0xbfffedd8) at
object.cc:127
#9  0x0046fbfe in Gtk::Object::destroy_ (this=0xbfffedd8) at
object.cc:260
#10 0x00173b15 in gtksourceview::SourceView::~SourceView
(this=0xbfffedd8, __in_chrg=<value optimized out>, __vtt_parm=<value
optimized out>) at sourceview.cc:445
#11 0x080490ff in main (a_argc=1, a_argv=0xbfffeee4) at basic/main.cc:30


> ---
> 3. Backport changes to 2.10 - when fixing completion stuff in 3.0 API I
> realized that this part is unusable in 2.0 - changes wrt. reference
> counting, completion test and conversions cleanup should be backported.
> And probably all TODO comments need to be removed.
> ---
> 
> ---
> 4. gtksourceview vs. Gsv - current namespace name (gtksourceview) is
> awkward and inconsistent with mm modules. I would rename it to Gsv. And
> maybe add a compat header file containing a line:
> 
> 'namespace gtksourceview = Gsv;'
> 
> What do you think?
> ---
> 
> 
> I'm not sure about the first two - maybe they are results of changes in
> Gtk+/gtkmm and needs to be corrected or maybe I'm just stupid and can't
> see obvious mistakes. Or maybe my jhbuild environment is screwed up.
> That way or another - I'm leaving it for a while (a month or two), since
> I have to do some other "away from programming keyboard" stuff, sorry.
> Feel free to push patches fixing above issues (maybe without 4th one) or
> file bugs in bugzilla with patches - I'll push them if they are all
> right.
> 
> Thanks,
> Krzesimir


-- 
murrayc murrayc com
www.murrayc.com
www.openismus.com