Criteria for GCC 4.0

[Joe Buck]

On Fri, Jun 04, 2004 at 12:54:28PM -0400, Andrew Pinski wrote:
> On Jun 4, 2004, at 12:48, Stephan T. Lavavej wrote:
> > Version number inflation is the natural order of things.  Better gcc 
> > 4.0 than binutils 2.15.91.0.1 (GAH).

> First that binutils number you mentioned is HJL's binutils and not the 
> GNU's binutils numbering system.

[ off-topic side rant ]
Yes, too bad all the GNU/Linux distros are using HJ's versions.  Nothing
against HJ, but if there's a problem with the FSF releases, I'd prefer it
if it were corrected.

[ back on topic ]

> The GNU's binutils is at 2.15.91 which means that will become 2.16 which
> is unlike gcc where we number our mainline CVS at the next major release
> number in this case 3.5.0 (or 4.0.0 for those other people).

We used to number GCC releases this way (there was a 3.2 and a 3.3).  But
this created confusion between discussion of the release branch or series
(which has two numbers) and release versions, so we decided to make it
3.4.0.  So now, x.y refers to a *series* of releases; the initial release
and the bug fix releases (3.4.1, etc).

On the 3.5 vs 4.0 discussion, I haven't seen anyone make a new argument in
a while; the same arguments are just being repeated.  Once the GCC Summit
is over, I think we'll just call an SC vote and be done with it.