[Top][All Lists]
[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: Applying outstanding patches [bug-grep]
From: |
Charles Levert |
Subject: |
Re: Applying outstanding patches [bug-grep] |
Date: |
Wed, 13 Apr 2005 22:00:01 -0400 |
User-agent: |
Mutt/1.4.1i |
* On Wednesday 2005-04-13 at 20:10:41 +0200, Claudio Fontana wrote:
>
> --- Julian Foad <address@hidden> wrote:
> > I have finally got around to spending some time on
> > Grep, and have started by
> > going through the issues in the Savannah tracker and
> > resolving those that are
> > easy to resolve. This includes applying patches
> > that are simple and/or
> > testable and that won't get in the way of applying
> > some of the larger patches
> > that are waiting.
> >
> > Stepan Kasal wrote: "Our main goal for grep 2.5.2 is
> > to get sane performance
> > with utf-8." While that is an important issue, my
> > main goal for the short term
> > is to fix as many correctness bugs as can be fixed
> > without making very large
> > (and therefore inherently dangerous) changes.
>
> I would suggest freezing the Red Hat UTF8 speedup
> patch. I studied it with attention and it addresses a
> real performance problem, but I think it needs some
> restructuring to avoid code dups, and generalize some
> functionality in appropriate functions to be more
> maintenable in the future.
Here's my 0.02 CAD.
I think that, given the current freeze we've
been in for some time, any motion (not even
necessarily proven progress) is good and we
should go ahead and apply the long-in-waiting
Red Hat patches.
Remember the obvious (IMHO):
-- CVS is supposed to be bleeding edge and
anybody who downloads it should know
what they're getting themselves into.
Users should look in the official releases
for stability.
-- It's even ok to break the CVS build, as long
as it's detected and fixed within a few
hours.
-- Things that build don't have to be perfect
the moment they enter CVS. There are
several ways to correct them afterwards:
-- by incremental improvements that build
upon the first commits;
-- by first backing out of commits that
turned out to have been unequivocal
mistakes after having received some
testing, and then applying newer cleaner
patches from scratch based on gained
experience.