Re: SUN make and gcc 3.0: committed changes to branch

[Phil Edwards <pedwards at disaster dot jaj dot com>]

On Wed, May 30, 2001 at 07:13:52PM +0200, Hans-Peter Nilsson wrote:
> The verdict was to just not fail.  See
> <URL:http://gcc.gnu.org/ml/gcc-patches/2001-05/msg02007.html>.

Ah, missed that one, thanks.


> >  There's not much point in checking for GNU
> > make if we don't use the resulting information.
> 
> Right, though supposedly one might want to use that information
> for maintainer-mode-like constructs, only emitted for GNU make;
> usable for developers.  See
> <URL:http://gcc.gnu.org/ml/gcc-patches/2001-05/msg01459.html>
> where I briefly noted that @ifGNUmake@ isn't used.

We've never used it.  When this test was added, a GNU extension was used
in (IIRC) src/Makefile.am for basic functionality.  Rather than making a
non-GNU-using person wait until the end of the build and then failing with
some weird error, we decided to do the test as early as possible, and then
tell the user which command name to use to insure that GNU make was employed.

Someone else pointed out that we don't check during the build itself that
the "make" found during configure is the same "make" now in use; all we
did was recommend a command.


> >  (Side question: do we
> > still require GNU make in the first place?)
> 
> If you mean that question as it stands in no particular context,
> then it was never an official requirement, now even less. :-)
> See discussion nearby URL above.

Not so much "officially," but more in the context of "what actually works?"
:-)  I haven't tried it myself in months.

I was about two minutes away from updating the libstdc++ FAQ and installation
instructions.  It'll be nice that this requirement can be dropped.


Phil

-- 
pedwards at disaster dot jaj dot com  |  pme at sources dot redhat dot com
devphil at several other less interesting addresses in various dot domains
The gods do not protect fools.  Fools are protected by more capable fools.