On 12/06/2011 04:16 PM, Daniel Richard G. wrote:
> Replacing a few parens with curly-braces on all platforms is ugly, but
> running an additional configure check on all platforms is nice?

Yes, since we already do other configure checks for make capabilities,
and substitute that into Makefile.in when producing Makefile.  And no
one said we have to run all the checks on all the platforms - it may be
sufficient to detect multiple features on a single make probe run, at
least for GNU make, to minimize forking in the common case.

> 
> What if the system has GNU Make and an old, broken make, and through
> some sort of PATH confusion, the configure-time check sees the former,
> and the user then invokes the latter? (Sure, the user should just use
> GNU Make, but why not avoid the possibility of such breakage in the
> first place?)

The user is already obligated to run the same make that was on the PATH
when they ran configure, or to run './configure MAKE=/path/to/make' to
hard-code their preference of make.  Automake already generates
Makefile.in that has substitutions that are implementation-specific; for
example, see how @SET_MAKE@, @am__include@, @am__quote@, and so forth
are hard-coded to the make implementation detected at configure time.

-- 
Eric Blake   eblake@redhat.com    +1-919-301-3266
Libvirt virtualization library http://libvirt.org