This is the mail archive of the gcc-patches@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: gcc/gcc ChangeLog doc/install.texi


Phil> Loren's comments are over here:
Phil> http://gcc.gnu.org/ml/libstdc++/2002-01/msg00182.html

Mark> I think Loren's #4 comment is very good.  I don't know if we actually
Mark> need to test for make having VPATH, but if we can it's a very good idea.

My point #4 does not actually advocate testing for VPATH support (nor
could we since at configure-time, we have no clue what make
implementation the user will invoke later).  Classically (i.e.  every
other FSF software package), ./configure does not require a make with
VPATH (top-level configure goes out of its way to ensure that a VPATH
make is not required in this configuration case).  Since gcc would be
the first to require it in this configuration case, I advocated a
warning to be emitted at the end of ./configure if Phil's patch was
taken.

I actually know of no make implementation that fails to implement
VPATH the way we like to use it.  Otherwise I would have tested Phil's
patch to see if it fails semi-gracefully and early with such an
implementation with Phil's patch.  Since I had no way to test it, I
just mentioned the issue.

Mark> I'd be all for using this patch (with the minor variants suggested
Mark> above by various folks) and then re-supporting the old build syntax.

My only goal for raising issues with Phil's patch were: (a) he asked
for critical feedback so I actually thought about the change a while;
and (b) I want to avoid a new class of PRs if we were to make a change
intended to squash another class of PRs... ;-)

Regards,
Loren


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