This is the mail archive of the gcc@gcc.gnu.org mailing list for the GCC project.


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]
Other format: [Raw text]

Re: patch for -Wno-long-long and early GNAT compilers


> Date: Mon, 18 Mar 2002 16:18:12 -0800
> From: Zack Weinberg <zack@codesourcery.com>
> To: Robert Dewar <dewar@gnat.com>
> Cc: kenner@vlsi1.ultra.nyu.edu, gcc@gcc.gnu.org

> From my point of view, what is considered to be a justifiable use of
> ACT's company resources is irrelevant.

I guess I fundamentally disagree with Zack here...  It doesn't serve
gcc to browbeat people into making them do things they don't want to
do.  Yes, it would be nice, and for some things, we should, but doing
so comes at the expense of non-contribution, which can be worse.

Would it be nice if Ada bootstrapped in a normal boostrap more often,
sure, should be browbeat them into making this happen, no.

If someone wants to submit contributions make it work on platform X,
then then can do so.  If no one cares to do this, then maybe it isn't
important.  The contention point, would be if people submitted these
contributions, and the Ada folks didn't want to put them in.  In that
case, we talk about about it on gcc with the major contributors to
gcc, and see if we can all agree on whether we do this or not.  If we
agree, done, if not, then it would be up to the SC.

I've reported bugs in gcc when other compilers (older unreleased
versions of gcc) were used to bootstrap it, and have been told, nope,
we aren't going to support this.  Did I loose sleep over it, nope,
just rebuilt a newbaseline compiler to bootstrap with.  It isn't a big
deal, really.

Remember, the most important contribution, is that they integrate it
with gcc 3.x, get the testsuite in and working, and test it, and
productize it; that they fix reported bugs in it...  What you want
seems to be so far down on the important list, as to fall off in my
mind.

I think the Ada folks can propose any limitation on what they can and
are willing to support and count on them for that.  They should be
granted a bit of latitude in this.  And if people are interested in
other goals that are not met that way, then they themselves can find
ways of making those goals happen.  Only if you are willing to have
them say no and accept that, is worthwhile to reject their
contribution.  I say, we decide if we want to tell them to take a hike
and go-away, if they are unwilling to make it work on pre 3.14p (or
whatever the technical case is).  We then inform them of the choice.
My vote is no, we don't so conditionalize they contribution.  An
appropriate use of this would be FSF copyrights, documenting code,
testcases for code, no regressions on all major platforms.

So, Zack, are you willing to tell them to go take a hike?


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]