[Xcb] [RFC PATCH] New XCB socket handoff mechanism for Xlib/XCB and other libraries

[Ian Osgood]

On Mar 15, 2008, at 11:40 PM, Barton C Massey wrote:

> In message <Pine.LNX.4.64.0803160709250.6522 at grozny.maths.univ- 
> evry.fr> you wrote:
>>>
>>> P.S. Somebody please help me finish the XCB port of
>>> x11perf?  I've got a working image library in util now, so
>>> it can't be that hard...  I'll get to it when I can, but if
>>> somebody would beat me to it that would be awesome.
>>>
>>> I really want to compare XCB performance to Xlib/XCB
>>> performance to find out what it's costing for what.
>>
>> would that port help me in finding why, now, evas with xcb is much  
>> slower
>> than xlib, using the new xcb_image ?
>
> Maybe.  I suspect that profiling should be sufficient,
> though, if the fault is xcb_image.  Probably need to add
> some fastpath code somewhere...
>
> 	Bart

Also, the image demos in xcb/demo/tests were all broken. I pushed  
fixes to at least make them compile. (What happened to  
XCB_ALL_PLANES?) Julia works, but lissajoux 0 and 1 are broken  
(xcb_image_create_native fails) and flames is broken (bus error,  
didn't delve).

xcbneko could also be adapted to the new bitmap functions at some point.

Ian