Re: GCC 4.0 vs. 3.5 -- don't decide now!

[Joe Buck <jbuck at welsh-buck dot org>]

Nathanael Nerode writes:

> Generally, version numbering seems to be about appearance as much as
> (or more than) anything else.

Well, of course.  Version numbers are about communication, or to use a
word that may be unpopular around here, about marketing.  That's why
your initial suggestion that sometime between stage 1 and stage 2,
some *technical* criterion would make it clear whether to call the
release GCC 3.5 or GCC 4.0, strikes me as odd.

Face it, it's marketing.  If we call it 4.0, the trade press (at least the
more clueful technical end, like lwn.net) will pay attention.  If we call it
3.5, no one will pay attention.  So the question is, is this a big enough
change that people should pay attention?

That's why, in the end, it's the SC's call; it's not technical, though some
technical arguments do enter into the decision.  Certainly it's a more
radical change in GCC internals than anything that was done between GCC
2 and GCC 3.  Ideally, if we want to make the splash that calling it 4.0
will cause, we should be able to show some performance wins, and we probably
will have some for C++ code (particularly in Boost).