CVS Update: xc (branch: trunk)

[Thomas Winischhofer]

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

Adam Jackson wrote:
> On Friday 01 July 2005 08:28, Thomas Winischhofer wrote:
> 
>>Zack Rusin wrote:
>>
>>>On Friday 01 July 2005 12:20, Thomas Winischhofer wrote:
>>>
>>>>PrepareCopy is defined getting two Bools, Reverse and UpsideDown.
>>>>However, in exa.c PerpareCopy gets passed two integers (xdir, ydir)
>>>>which are >= 0 for "left to right"/"top to bottom".
>>>
>>>Ah, yeah, missed it. Fixed. Thanks :)
>>>
>>>Zack
>>
>>Wow... this thing is superfast for transparency.... even without
>>Composite acceleration. Good work, folks!
>>
>>I have already ported the sis driver entirely to exa (including
>>adaptions for dga and xv), will commit as soon as I had the chance to
>>test it intensively on all my supported chipsets.
>>
>>Feature freeze was on Monday, wasn't it...? Hm, looks like a weekend by
>>the computer...
> 
> 
> Will be on Monday (July 4).  Future tense.  However, we're expecting to allow 
> exa additions to drivers for a while yet, with the expectation that the exa 
> support will be built by default but not necessarily enabled by default.
> 
> There's some ambiguity here in the xorg.conf option naming, so here's what I 
> think we should do:
> 
> Option "Accel" "<bool>": Enables hardware acceleration.  Defaults to on.
> 
> Option "AccelMethod" "[exa|xaa|...]": Selects the acceleration architecture 
> the driver will use.  Defaults to "xaa".
> 
> That last part might be contentious.  The logic here is that while exa is new 
> and sexy, it's not nearly as well tested as xaa, which sucks but at least 
> sucks reliably.  I'm willing to defer to maintainer opinion for those drivers 
> that have strong maintainership - sis, via, voodoo, etc - but in general I 
> would prefer that xaa still be the default for 7.0.

I already have done all this. The option (its name is ok with me) is
only supported as long as both XAA and EXA support are compiled into the
driver. The sis driver is, in the meantime, completely "modularized", ie
removing XAA is done by #undef'ing a symbol.

By default, XAA is used. I completely agree that exa first needs more
testing. (Done some tests the last 6 hours, it seems totally stable, but
who knows what corner cases I haven't reached yet).

Thomas

- --
Thomas Winischhofer
Vienna/Austria
thomas AT winischhofer DOT net          http://www.winischhofer.net/
twini AT xfree86 DOT org
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.1 (GNU/Linux)
Comment: Using GnuPG with Thunderbird - http://enigmail.mozdev.org

iD8DBQFCxYYBzydIRAktyUcRAisWAJ9u7zY2yxgZBwyBaQlKO1NV5fXBigCglEPl
dd3Qcj9YTnYHtOHM/1Bbtyw=
=lJzi
-----END PGP SIGNATURE-----