Release Criteria (Was Re: Radical proposal: skip 3.4)

[Andrew Pinski <pinskia at physics dot uc dot edu>]

On Jan 10, 2004, at 16:06, Michael Elizabeth Chastain wrote:

> Steven Bosscher wrote:
> > Of course the major regressions are very serious.  But regressions
> > like 12978, 13376, 12835, 13577, 12267, and so on, are really minor,
> > more like enhancement.

Note out that list 12978 have a patch and most likely 12267 is fixed.

> I strongly disagree with your priority assessment of 12267.
>
> 12267 makes C++ with stabs+ undebuggable.  If gcc 3.4 ships and 12267 
> is
> still present, I'm going to recommend that people with stabs+ platforms
> either stay with gcc 3.3 or switch to dwarf-2, and propose language to
> that effect for the gdb release notes.  Some of them will choose to 
> stay
> with gcc 3.3 and that will make my life harder as a gdb tester.

Just to get this clear 12267 is about debugging C with stabs and not 
C++ with stabs.
I think you are talking about PR 13539 which is about protected 
inheritance and stabs+
which I posted the patch already, I just need get the approval and I 
will commit it, I
really want this one fixed too because I use stabs as I am on a target 
which does not
support dwarf2 at this moment (this will change but it takes time to 
get support in
the as and linker and in the executable format).

As a separate issue, us developers really need to know what the Release 
Criteria is
for 3.4 because there is none as of yet, and we really need to know 
what bugs really
should be fixed for 3.4 and what can be delayed, for an example I would 
consider
some diagnostic problems as not branch blocking (unless there is a 
patch already to
fix it).  Also what is the platform support do we want for 3.4, the 
same as 3.3 or
updated versions of some, like AIX 5.0 instead of AIX 4.3.3.0, Solaris 
2.9 instead of
2.8. Maybe we also want to move x86_64 up?

Also should there ABI bugs be fixed once and for all for 3.4 or should 
they not be
targeted at all?

What about the applications for 3.4?
What Linux version?
What QT version?

I know that have been some recent (aka today) some effects to reduce 
the memory
and compile time problems in GCC. So should we also add some of those 
PRs to our
Criteria and have them explicit for 3.5 also (planning ahead)?

For the next release after 3.4 (what ever it is going to be called) we 
should think of
these right after 3.4 gets released so we do not have this confusion 
before going into
stage3.

Steven and I have moved some bugs which we feel that are not important 
to be fixed for
3.4.0 to be target either 3.4.1 or 3.5.0.  Also the bugs marked as 
critical are the
ones which should hold up any release of the compiler and not 
necessarily the ones
targeted to 3.4.0.

Thanks,
Andrew Pinski