This is the mail archive of the
gcc@gcc.gnu.org
mailing list for the GCC project.
Re: Beyond GCC 3.0
- To: gcc at gcc dot gnu dot org
- Subject: Re: Beyond GCC 3.0
- From: Marc Espie <espie at quatramaran dot ens dot fr>
- Date: Sun, 1 Jul 2001 14:25:42 +0200
- Organization: Ecole Normale Superieure (quatramaran)
First, I really hope this does happen. Having a sound release schedule
is very, very cool. All people in the OpenBSD project will back you on
this one.
A few experienced points I have:
- six months may look too short, but it's feasible. I don't think a Unix
kernel is less complex than gcc, and we do manage to pull it out every six
months.
- sometimes, critical bugs occur. Sometimes, the fix is really simple.
Patches can be released between actual releases. They're less heavy-weight,
and they help tremendously.
- one way to help people concentrate on release issues is to FREEZE everything
else... during the last development step, disallow ANYTHING to happen on the
cvs repository that isn't release-related... you find out that people are
much more motivated to fix bugs when they aren't allowed to commit anything
else until the bugs are fixed... You may say `oh, this is an open project,
if we keep developers out, they will go elsewhere.' Well, just try it, it
works better than that.
I've seen a few comments about major/minor releases. Well, I think that
point is moot. In a `correct' release process, features are introduced,
tested, and released when they're ready, at which point you start worrying
whether what you are creating is a new `major' release or a new `minor'
release. Basically, what's wanted is an `up' path, where more or less each
new release of gcc is better than the last (more stable, faster, more stable,
better code, more stable, more features, more stable.... see a trend ?)
So far, this would be a LARGE improvement about gcc's track record over the
last few years, where the switch from a version of gcc to the next version
has always been painful for people down the chain.
Well, I don't want to end this on harsh words. I'm downright happy with the
way things are going with gcc right now. Due to personal, familial
circumstances, I haven't had nearly enough time to deal with gcc and
OpenBSD recently---which means that gcc 3.0 doesn't even bootstrap without
patches right now, but from what I've seen in lurking on the mailing-lists,
the focus is changing towards the right direction. Kuddoes to all.