Re: abort in gmessages.c

[Owen Taylor <otaylor redhat com>]

On Wed, 2003-05-21 at 11:58, Emmanuel Briot wrote:
> On Wed, 2003-05-21 at 17:52, Owen Taylor wrote:
> > What we try to do in such cases is detect the problems early
> > enough that we can harmlessly return and leave things in
> > a sane state, with a g_warning() to alert the programmer
> > that that there is a bug they need to fix.
> > 
> > Can you patch GtkTextBuffer to do that in this case?
> 
> We are certainly puting in place some tests that the (line, column) pair
> is valid. 
> I am not sure however why the gtktextbuffer would raise a fatal error in
> such a case. It could print a warning and then default to the last
> character on the line for instance. That would seem more user-friendly,
> I believe.

You really aren't giving me enough information to comment meaningfully
here :-)

 - What error are you triggering in particular?
 - What is the minimal code to trigger this g_error()?

Most g_error() uses in the GtkTextBuffer code are for internal
consistency failures.

Regards,
                                          Owen