On Thu, Mar 29, 2001 at 12:41:15PM -0800, Jeffrey W. Baker wrote:
> At some stages in time, some things are going to be broken, but nobody is
> forcing the user to run the latest and greatest. CVS has the ability to
> checkout arbitrary dates and revisions. Take advantage of that.
FWIW, I have been trying various CVS dates.. This leads to conclusion
that it's not some recent commit causing a nasty problem. I've been the
victim of (and occasionally the cause of) such with projects in CVS
before.
Anyone who remembers QuakeForge's initial autoconf support will attest
that I worked all day and night on it, finally getting it to work, then
went to bed after committing it. Woke up to find it reverted because not
only did it only work on my system, but nobody else could make heads or
tails of it to work around the problem or even branch it properly. I
fixed it, but until then nobody else was crazy enough to look at it.
In this case, it's starting to look like disabling the AGP stuff
completely did the trick. Most of the time it's been running has been
spent in Xawtv fullscreen which is hardly a stress test (I think?) but
things like mode changes and apps like netscape tended to significantly
reduce life expectency of my processes. If that's the case, I was
completely wrong about the radeon driver. I understand it'd also mean
I've got a bigger problem and it wouldn't matter which card I was using
until the bug is found and splattered. Wonderful.
--
Joseph Carter <[EMAIL PROTECTED]> Free software developer
<Deek> Exactly how much of a PITA is this in C?
<Knghtbrd> It's written in C++.
<Deek> Hence my question.
<Knghtbrd> I could do something like it in C. Anyone who saw the results
would think I was either a genius or out of my fucking mind.
They'd be right on either count.
PGP signature
