[prev in list] [next in list] [prev in thread] [next in thread]
List: freedesktop-xorg
Subject: Re: render improvements
From: Lars Knoll <lars () trolltech ! com>
Date: 2005-06-15 20:53:44
Message-ID: 200506152253.44882.lars () trolltech ! com
[Download RAW message or body]
On Wednesday 15 June 2005 21:57, Keith Packard wrote:
> On Wed, 2005-06-15 at 20:32 +0200, Lars Knoll wrote:
> > I'll try to convince him :)
>
> You don't have to convince him, just make sure your own emacs
> configuration can be set to leave my X server files alone :-)
>
> > The problem is that both inside Trolltech and KDE it is the default to
> > remove white space at end of lines when saving. I'll see if I can remove
> > the hook that does this for C code.
>
> This policy seems misguided to me, but then I spend a lot of time lookin
> at CVS diffs, for which changes of this nature are evil.
Well, as long as everyone uses this, you won't see any problems. That's the
case for KDE and Qt. As I said, I'll make sure my emacs leaves them alone for
X.
> > As a reference I attached the patch without the white space changes. It
> > basically correctly implements projective transformations (I
> > misunderstood this and only implemented affine ones before), fixes some
> > bugs in the transformation code and has a small fix in the convolution
> > code.
>
> Ah, note that the provided diff is also hard to read because there are
> some *important* whitespace changes that it ignores. That's just as bad
> as including white space changes which aren't important. I refer here
> to whitespace changes which affect how the code is indented.
True, but it's at least a bit better.
> > I also expanded some code to get somewhat smaller and faster inner loops.
> > This is not very relevant at the moment as the framebuffer reads are
> > limiting us, but in the long run with DMA support it will probably start
> > to matter.
>
> Even in the short term we should be able to get the x server to be a lot
> smarter about where pixmaps get put so that we don't spend a lot of CPU
> cycles waiting for pixels to trickle across the AGP bus.
Yes. I've just seem something today that looks very promising here :)
> Thanks again.
You're welcome. I hope I the gradient support (this time hopefully without
whitespace issues) ready in time for the conference.
Cheers,
Lars
[prev in list] [next in list] [prev in thread] [next in thread]
Configure |
About |
News |
Add a list |
Sponsored by KoreLogic