[gtk-list] Re: announce: yet another gtk+ C++ wrapper (no caps as

[Mario Motta <mmotta guest net>]

On Fri, 11 Sep 1998, Mario Motta wrote:
>On Fri, 11 Sep 1998, Karl Nelson wrote:
>>
>
>>...  (can anyone tell me how the signal frame 
>>differs from VDK to Gtk--?) 
>
>In few word the VDK side: VDK uses a static early binding template table where
>it looks up at run time. Tables are inherited down the hierarchy and look up
>algorithm is a straight teta(n) recursive one. Trivial benchmarks on my P 200
>Mhz showed a teta(1) performance in almost all cases, (roughly about  500
>signals/events catched, mapped and answered per sec but i should investigate
>more..) Speed and simplicity should and have to go togheter, that's a target for
>me.
>Gtk-- side: i guess that follows Gtk+ route binding  member function offset
>address with real run-time object address in order to have the correct absolute
>address at runtime. But gtk-- makers can know better than me the internals.
>>
>>--Karl
>friendly
>
i would just point out that a less than trivial benchmark about VDK speed in
dispatching signals showed roughly 1000 signals dispatched and answered
per sec. This test was conducted in P 200 Mhz with not optimized, off-line
inline functions code under debugger. Optimized code should be much faster.
Time rate of growth is linearly teta(n) dependent on number on signals
connected to a single form (window in Gtk+) , in practice if n < 50 can be
considered constant mainly because gtk_signal_emit_by_name() call overhead  
should be prevalent.
regards.
 --
Mario Motta
===========
AI Research Group - Rimini
mmotta@guest.net
http://www.guest.net/homepages/mmotta/index.htm