Re: Bootstrap status of 2.95.3test2 and 3.0 on many host types

[Gerald Pfeifer <pfeifer at dbai dot tuwien dot ac dot at>]

On Sat, 27 Jan 2001, Martin Buchholz wrote:
> gcc 2.95.test3
>
> 2.95/i386-unknown-freebsdelf	bootstrap successful
>
> gcc 3.0 CVS of 2001-01-26:
>
> 3.0/i386-unknown-freebsd4.2	make tries to run `+make' !?
>
> 2.95, but not 3.0, builds on Freebsd 4.2.  It looks like 3.0 may have
> a dependency on GNU make or be confused by `set -x'.

Yes, it seems that there are GNU make dependencies (and possibly also a
bug in FreeBSD make), according to discussions earlier this week.

As it did work for GCC 2.95, I'd say this is a regression with current
CVS sources of GCC 3.0. A patch has been submitted for FreeBSD make that
might fix a possibly related bug; once that is in, we should test again.

Not that I am trying to push FreeBSD make, but I believe that, where
multiple implementations exist that implement some standard, POSIX in
this case, even though GCC seems to use some commen extensions according
to Zack, we should regularily try to use more than one. Portability is
a quality of implementation issue, IMHO.

Gerald
-- 
Gerald "Jerry" pfeifer@dbai.tuwien.ac.at http://www.dbai.tuwien.ac.at/~pfeifer/