Re: catching exceptions

[muppet <scott asofyet org>]

On Apr 30, 2008, at 6:47 AM, Gabor Szabo wrote:

> Hi,
>
> On Wed, Apr 30, 2008 at 8:04 AM, muppet <scott asofyet org> wrote:
> > On Apr 29, 2008, at 9:48 AM, Gabor Szabo wrote:
> >
> > > Can I catch this exception somehow?
> >
> > With some work, yes, you can trap it with a perl handler.  This is  
> > not
> > really generic, though, and is not a proper solution to your problem.
>
> As a Perl programmer I think I should be able to trap every  
> exception, even if that occurs in the C-level code beneath.
>
> I think it would be better if the Perl binding would trap these and  
> throw them as exceptions in the Perl code.
>
> Please consider adding this to future releases of Gtk2.


While i agree that a perl programmer should be able to trap all  
exceptions, there is a technical problem in the case of fatal g_log()s  
(such as g_error()) --- C code calling g_error() expects that the  
program will abort().  Program state is undefined after trapping such  
an error, and there's no way for the bindings to clean up and prevent  
crashes.  In many cases you'll get lucky with a longjmp() and only  
have a leak, but it's simply not safe.



> Would it be possible to add the above convenience function to the  
> standard
> Gtk2 Perl distribution?

We try to avoid such additions in the name of API sanity.  For  
example, if gtk+ later added a method with the same name but slightly  
different semantics, we would have a name clash, and would have to do  
very ugly things (for both binders and users) to resolve that.


--
One, two, free, four, five, six, sebben, eight, nine, ten, elebben,  
twull, fourteen, sickteen, sebbenteen, eightteen, elebbenteen,  
fiffeen, elebbenteen!
  -- Zella, aged three, counting to twenty.