Another patch for egcs 1.1.2

Mark Mitchell
Sun Feb 28 18:15:00 GMT 1999

I think we (perhaps I mean Jeff, rather than some amorphous we?) had
better come up with some firm guidelines for what kind of things go
into patch releases.

The following seems to happen every time we get near a release:

  o We mostly freeze the release.
  o HJ and others submit lots of patches.
  o We get stuck, and iterate for ages.

Meanwhile, the mainline, which has made tremendous progress, gets
farther from the light of day while we worry about the patches.  I,
for one, would like to see 1.1.2 out, and move on to 1.2, which has
better optimizations, better reliability, a much improved C++
front-end, and doubtless lots of other features I don't even know

There are literally tens, if not hundreds, of substantial C++ fixes
that we have made since the 1.1 branch.  There are lots of people
reporting C++ bugs in 1.1.1 that are fixed on the mainline.  We could
move all of the changes to the 1.1.2 release, but I don't think that
makes sense.  Instead, they'll just have to be disappointed; C++
support in 1.1.2 is going to be just like that in 1.1.1.  In 1.2,
they'll be in for a treat; lots of bugs fixed.

It's not that the patches are bad; the point is just that we can't
make 1.1.2 perfect.  Instead, IMO, 1.1.2 should be an emergency
absolute show-stopper bug-fix release.  We should be moving on to 1.2,
and try to make that release as good as possible.  We don't need
Jeff's time, our tester's time, and our focus distracted from 1.2.

Mark Mitchell
Mark Mitchell Consulting

More information about the Gcc-patches mailing list