kenner@lab.ultra.nyu.edu (Richard Kenner) writes:

> In article <5uqh3g$1to$1@bbj.freenix.fr> card@bbj.freenix.fr (Remy Card) writes:
> 
> >Hmmm, when I read the "Tired of waiting..." thread, I got the
> >impression that people working on the g++ part were quite frustrated
> >that a new release of gcc/g++ did not happen, because they had much a better
> >c++ compiler and they had to wait for the FSF gcc maintainers to say "Ok,
> >it's time to release a new version of gcc/g++", even if they got reports
> >for bugs that they had fixed one year ago...
> 
> That doesn't make much sense for two reasons:
> 
> (1) The startup of the EGCS project is far more likely to slow down
> the final testing of GCC 2.8.0 (and hence release) than speed it up
> since the people who would otherwise be working on that task are now
> splitting their efforts between that and EGCS.

A year seems a somewhat long time for "final testing", so I don't
think that final testing of gcc 2.8.0 has been a full-time occupation
for all of the involved parties.

> (2) The EGCS project is just starting up and the 2.8 cycle is coming
> to an end.  The goals of the EGCS project are to things *after* the
> 2.8 cycle, not of that cycle itself.

Mostly yes, because actually nobody can believe that 2.8.0 will be
long coming *now* (but most people believed this already a year ago).
What the EGCS project wants to avoid, probably, is that bug reports
and development focus on outdated work, like it has happened with
2.7-derived compilers, particularly with the g++ frontend.  2.7 is
more than 2 years old, and at least for the last year this has grown
into a major nuisance for C++ users, as the need for a more correct
compiler got more and more pressing for many people.

Personally, I can't actually share your opinion that your arguments
show that trying to have people improve and test the actual streak of
development instead of that from two years ago does not make much
sense.

> >Are you kidding?  Can you seriously say "a few months" when the
> >last release of GCC is 2 years old? :-)
> 
> Yes.  The point is that had EGCS been around two years ago, the
> work that we held up GCC 2.8 for would have been shunted off to it
> and we would have been at 2.10 by now.

So you think that EGCS actually is a good idea, as we would have more
features, and better-tested than we have now.

> 
> >I don't know anything about compilers, but I like how egcs is
> >developped and made available to testers on a regular basis.  If the egcs
> >developpers happen to have a version that is stable enough to be available
> >as a release, I will certainly use it and not use the FSF gcc anymore.
> 
> There's some confusion here.  Once a change has been shown stable
> enough within the EGCS framework, it gets put into the mainstream
> and EGCS goes on to test *other* unstable changes.

Let's see whether this works out.  I doubt it.  Quite a few changes
have been shown stable enough in the Cygnus development framework, yet
they still have not appeared in the "mainstream".


-- 
David Kastrup                                     Phone: +49-234-700-5570
Email: dak@neuroinformatik.ruhr-uni-bochum.de       Fax: +49-234-709-4209
Institut f�r Neuroinformatik, Universit�tsstr. 150, 44780 Bochum, Germany