Follow-up Comment #7, bug #48643 (project make):
Perhaps your argument is (similar to Boris's argument in bug #17752 ) that the
algorithm in the manual is bad and should be changed; in particular the rule
5.3 that says "ought to exist" means "mentioned as a target or a prerequisite"
anywhere in the makefile is not right.
That may be so, but I don't think the change you've proposed here is
sufficient, similar to that conversation we had back in 2009.
The change you suggest, so far as I understand, changes the 5.3 rule so that
"ought to exist" means that the file must actually be mentioned as a target,
and being mentioned as a prerequisite is not sufficient.
At the very least, if we decide to move forward with this, we have to install
a documentation change and mention it in the NEWS file as a
backward-compatibility break.
The problem with this change is that it does have other effects. For one
thing, it causes some regression tests related to intermediate files to fail.
That's because after this change, targets that were previously considered as
"ought to exist" are no longer considered so, and thus where previously we
would stop at step #5 now we fall through to step #6 which changes the
behavior.
_______________________________________________________
Reply to this item at:
<https://savannah.gnu.org/bugs/?48643>
_______________________________________________
Message sent via Savannah
https://savannah.gnu.org/
