Subject: Commit: back out martin's changes ;-(
From: Dom Lachowicz (dominicl@seas.upenn.edu)
Date: Wed Sep 05 2001 - 09:44:05 CDT
I'm really sorry that I'm doing this, but it needs to be done.
1) KWord import is entirely non-functional. Export is barely functional, but
I'm willing to let that much slide. I'm working with Christian Meyer and
Kenneth Christiansen so that they get comfortable enough with the codebase and
C++ in general so that they can begin working on it again.
2) UT_Timer: I should have spoken up sooner on this issue, but I wasn't sure
that anyone was going to commit this patch. I guess that I was wrong. The
patch itself is fine for what it is. The proposed solution to the real problem
is entirely wrong, however. In any event, this shouldn't be committed a few
hours before the tree is closing, completely sans-testing on non-UNIX (and
without any real testing on UNIX for that matter). IMO we're introducing a
major design flaw into the tree. The timers themselves are enough of a hack
(ask Stephane Frisch - Be guy - about that sometime). Idle queues are a
grosser hack on top of that. The results, however, are real, and I won't deny
that much.
If you want to persuade me that this patch should be committed (which will be
an uphill battle for you), you will need to do it after the tree closes and we
branch. If you want me to design an XP thread+mutex class or evaluate the ones
that already exist, than that's fine too. I can handle this, and it is a
*much* better solution, tho it will require much more work on our part.
I'm sorry for stepping on people's toes.
Dom
/tree Nazi
This archive was generated by hypermail 2b25 : Wed Sep 05 2001 - 09:44:10 CDT