Re: GCC 3.0 Release Criteria

[Andreas Jaeger <aj at suse dot de>]

>>>>> Mark Mitchell writes:

Mark> The original list did not contain Intel x86 Debian GNU/Linux, but this
Mark> platform was suggested strongly because it is, as I understand it, the
Mark> official GNU distribution.  RedHat Linux appears twice because I
Mark> believe it to be the most widely used distribution of Linux; that
Mark> means that failing to work correctly there will affect a lot of
Mark> people, that a lot of people can provide testing cycles, and that a
Mark> lot of people are incentivized to help fix bugs.

Does it really make sense to test for both RedHat and Debian at all?
What's the difference for gcc between different distributions?  AFAIK:
- glibc version
- kernel headers
- sed,awk,bash,make,... versions

There shouldn't be much difference between *recent* Linux
distributions [1].

IMO we should test instead just for Linux - and perhaps specify
somehow the versions of glibc, kernel [3] we want to test - or mention
that we do expect recent versions of these.  This would also give
people with self compiled distributions the chance to make an impact.

It would make sense to test for totally different glibc or Linux
versions but e.g. testing with glibc 2.0.6 and 2.1.3 wouldn't make
sense since you can't compile glibc 2.0.6 with gcc 2.96 at all[2] (the
same is true for Linux 2.0.x.)

Andreas

Footnotes: 
[1]  For example and AFAIK RedHat, Debian and SuSE all use glibc 2.1.3
     and Linux 2.2.14 (sometimes with different drivers) - but essentially the same.
[2]  glibc 2.0.6 contains some assembler functions gcc 2.96 will not
     accept - and I'm not sure about strict-aliasing.
[3]  Or whatever seems relevant.
-- 
 Andreas Jaeger
  SuSE Labs aj@suse.de
   private aj@arthur.rhein-neckar.de