[haiku-appserver] Re: accelerating app_server

  • From: Ingo Weinhold <bonefish@xxxxxxxxxxxxxxx>
  • To: haiku-appserver@xxxxxxxxxxxxx
  • Date: Sun, 22 Jul 2007 14:40:27 +0200

On 2007-07-22 at 00:04:54 [+0200], Stephan Assmus <superstippi@xxxxxx> 
> Ingo Weinhold wrote (2007-07-21, 23:39:33 [+0200]):
> > 
> > On 2007-07-21 at 23:05:40 [+0200], Stephan Assmus <superstippi@xxxxxx>
> > wrote: [...]
> > > Well... I just felt I write this email and see if someone has some
> > > ideas. :-)
> > 
> > For the client side you can use the strace command to see the syscalls
> > being used. This would at least help to analyze whether our
> > implementation uses more syscalls. This should also answer your question
> > regarding whether Dano communicates with the app server when the drawing
> > is fully out of the clipping region.
> I think I figured that one out already. The times for just the StrokeLine
> calls are exactly the same for drawing inside and outside the clipping.
> When I run both the Dano version and the testenvironment version of my
> program in strace, it seems like neither of them is using syscalls for
> communicating with the respective app_server. I get a lot of output when I
> launch or quit the programs, as well as on focus changes, but not during
> the redraws. Am I doing something wrong?

Maybe you didn't pass the "-T" option to strace? Since the drawing code 
likely runs in another thread, you wouldn't get any output. Not sure why 
the app thread would be involved in focus changes, though.

CU, Ingo

Other related posts: