Re: [HELP REQUEST] property notifies and coalescing them

[Tim Janik <timj gtk org>]

On Fri, 10 Aug 2001, Hans Breuer wrote:

> At 17:11 10.08.01 +0200, Tim Janik wrote:
> >hi all,
> >
> >could people, when adding property notification, please
> >make an attempt at coalescing them where possible?
> >
> >the only place where this is not required is inside of _set_property()
> >implementations. for other functions that call g_object_notify() multiple
> >times, or funcitons that call multiple subroutines which notify
> >properties, adding g_object_freeze_notify/g_object_thaw_notify
> >is appropriate.
> >
> >if some volounteer has time to investigate multiple property
> >notifies without freezing/thawing the queue and missing property
> >notification in general (that is, a property ::foo being changed on
> >an object without ::foo being notified, e.g. in a object_set_foo()
> >function), that would be apprechiated.
> >
> This sounds like being another simple perfomance booster. So I volunteer
> to do it, because I appear to use the slowest computer :-)
> 
> BTW: (from one of my 'Re: some performance notes')
> > * ~10% of the non-debug time was spent copying GdkEvent structures.
> >   This is easily fixed by adding G_SIGNAL_STATIC_SCOPE to
> >   the GdkEvent signals. (this was part of the signal overhead
> >   mentioned above)
> >
> This is adding '| G_SIGNAL_TYPE_STATIC_SCOPE' to every 
> g_new_signal (..., GDK_TYPE_EVENT), right ?
> It appears to really give a performance gain of about 10%.
> Ok to apply a patch ?

at least for all events in gtkwidget.c, which are emitted from
gtk_widget_event(), this is ok, as the event will definitely
stay around for the duration of a gtk_widget_event() call (the
caller has to ensure that).

>  
> Thanks,
> 	Hans

---
ciaoTJ