Re: Beyond GCC 3.0: Summing Up

[Stan Shebs <shebs at apple dot com>]

"Joseph S. Myers" wrote:
> 
> On Tue, 10 Jul 2001, Stan Shebs wrote:
> 
> > GDB already has a two-stage process for this, where the first stage
> > is to research the status of the target, and if it appears moribund,
> > declare its support obsolete, comment out all the code, and make a
> > release with the obsolete code still present.  Then if nobody speaks
> > to save it, delete it from the next release.  So far none of the
> > obsoleted targets has had to be unexpectedly resurrected.
> 
> Given the use of CVS, what is the advantage of this process over simply
> announcing on gcc-announce the targets it is proposed to remove and seeing
> if anyone objects, as in
> <URL:http://gcc.gnu.org/ml/gcc-announce/2000/msg00002.html>?

Not everybody who uses GCC is on these lists - you'd be surprised
how far afield GNU gets.  Several years ago, when I was maintaining
GDB, the FSF forwarded me a bug report in paper mail that they'd
received from a university deep in the interior of China, complete
with a daisywheel printout of some of the stab strings in a Fortran
program.  (Unfortunately, I never did make sense of the report.)

The GDB process is a little more careful, since CVS can get corrupted
and old data become irrecoverable, as people here have discovered,
while releases are mirrored in many places around the world and
are easily unpacked.  But in truth it probably doesn't matter a
whole lot, since the configs are dead.  (Although overeager and
undereducated GDBers have a couple times recently proposed whacking
live configs, fortunately somebody spoke up right away.)

Stan