[g-a-devel] Re: gail - gdk lock take/release ...

[michael meeks <michael meeks novell com>]

On Tue, 2005-11-22 at 13:35 +0000, Bill Haneman wrote:
> Thanks for the patch.  The strcmp bug has been fixed for days, are you 
> sure you patched HEAD?

	Ah - you're quite right - I did diff head; as you can see the change is
not in the patch, only ChangeLog ;-) noticed it before update.

> Do you want me to review the patch as-is, or refactor the cut and paste 
> stuff and apply myself?

	As you like. I think there is some serious benefit to a gail_add_idle()
call - to shrink this patch by taking / releasing the GDK_THREADS mutex
as appropriate, of course, simplifying the change to the code & also
dramatically reducing the probability of error in the cut & paste if (!
widget || !is_sensitive || !is_visible) stuff.

	Also - of course, storing a closure & doing a g_signal_connect
("destroy", ... ) which would (optionally) de-register the idle handler
on destroy would clean out a load of code & help centralized that for
further potential locking fixes.

	So as you like really :-) of course, there is some merit & little harm
in committing the existing patch,

	HTH,

		Michael.

-- 
 michael meeks novell com  <><, Pseudo Engineer, itinerant idiot