Re: RedHat 7.0 + gcc-2.96, what's going on with gcc releases?

[Mike Stump <mrs at windriver dot com>]

> Date: Tue, 01 Aug 2000 18:39:54 +0200
> From: Franz Sirl <Franz.Sirl-kernel@lauterbach.com>

> the just released rh70 beta contains a gcc-2.96 snapshot as the main
> compiler it seems.

Ok.

> Since gcc-2.96/gcc-3.0 is nowhere near a release AFAIK, I would like
> to know if it makes sense to drop the planned (?) gcc-2.95.3 in
> favor of a gcc-2.96 release with the libstdc++-v3 and new-abi stuff
> still disabled?

No.

> I for myself think this makes sense, cause as soon as rh70 is out,
> there will probably more bug reports against the snapshot that RH
> uses than for gcc-2.95.x simply due the large userbase of RH.

So?  I don't see a problem as those bugs reports are more relevant
than bug reports against 2.95.2, if our quality was up high enough in
their release.  They do have the resources to test and assure quality,
and I expect they use them, so I don't see a problem.


I consider it unfortunate, but that is the risk we take of not serving
our users.  RedHat happens to be a user.


I am in the same situation.  I needed a certain feature in the
compiler, it was only in a newer compiler, and I developed and shipped
a 2.96 era compiler to my customers.  Life goes on.  And next week (or
next month), I'll do it again, for all the same reasons all over
again.

This is SOP.  I don't see it going away real soon.

Anyway, my point is, just because a company needs to do this from time
to time, doesn't mean we should contort what we are doing or take
immediate action because of it.  We should fold it in and if we think
we can make changes that lessen the need for other companies to do
this, at a cost that we find reasonable, then we should do it.


The maybe more interesting point, would be the shared library
numbering scheme, C++ binary compatibility issues and so on.