Re: ansi5.test fix

[Alexandre Duret-Lutz]

>>> "Tom" == Tom Tromey <address@hidden> writes:

[...]

 adl> I think using
 adl> U=_ $MAKE -e
 adl> would be better, but I'm wondering if there are some
 adl> drawbacks with using -e.

 Tom> I don't know.

I've just sent an update of the "Limitations of Make" section of
the Autoconf to a few lists for comments.  This include some
paragraphs about `make -e'.  I guess someone will react if this
is bad, so I'll wait a few days before committing this patch.

 adl> The patch below updates maintainer-check to detect occurrences of
 adl> "$MAKE foo=bar" and suggest "foo=bar $MAKE -e".

 Tom> This only matters if we have subdirs, right?

I think it also matters for a few other targets, like
install-strip.  Any target that calls $(MAKE).

 Tom> Anyway, I don't have a real problem with this.
 Tom> It might be more reliable to simply sed `U = _' into the Makefiles.
 Tom> But I'm not too concerned about it.

I'll do this if s.o. says `make -e' is bad.  Otherwise I'm quite
happy to put a few `make -e' in the testsuite, until someone
complain so we learn the conditions under which `make -e' is not
reliable.
-- 
Alexandre Duret-Lutz