wow. if this is the support that you get from free software imagine
the quality of support that we must get from cadence or synopsys or
avanti! that must be fantastic!
Paul D. Smith writes:
> %% "Albert L. Ting" <[EMAIL PROTECTED]> writes:
>
> >> RCS
> >> Any file `N' is extracted if necessary from an RCS file named
> >> either `N,v' or `RCS/N,v'. `N' will not be extracted from RCS if it
> >> already exists, even if the RCS file is newer.
>
> alt> The above rule, where the RCS file is newer, does not seem to
> alt> work if the location of the RCS'd file is in a directory defined
> alt> by VPATH. Enclosed is a tar file but I'll explain the test case.
> alt> The directory structure is:
>
> alt> drwxrwx--- alt 0 2000-03-17 17:01 make-bug2/
> alt> -rw-rw---- alt 79 2000-03-17 17:01 make-bug2/Makefile
> alt> drwxrwx--- alt 0 2000-03-17 17:00 make-bug2/foo/
> alt> drwxrwx--- alt 0 2000-03-17 17:00 make-bug2/foo/RCS/
> alt> -r--r----- alt 184 2000-03-17 16:40 make-bug2/foo/RCS/bar.old,v
> alt> -r--r----- alt 8 1999-06-07 14:29 make-bug2/foo/bar.old
>
> alt> The makefile is:
>
> alt> VPATH := foo
> alt> foo: bar.new
> alt> %.new: %.old
> alt> cp $^ $@
>
> alt> Typing "make" will "co" a copy of bar.old into the directory,
> alt> which it shouldn't.
>
> Hmm. I disagree that this is a bug.
>
> The rule you cite isn't implemented in make per se; that is, there's
> nothing in the target/prerequisite setup that makes it true.
>
> What makes it true is that the shell script for the checkout tests to
> see if the file exists and if so, it doesn't actually _do_ the checkout.
>
> When VPATH enters the mix, consider this statement from the VPATH
> documentation "How Directory Searches Are Performed":
>
> 4. After processing the prerequisites, the target may or may not need
> to be rebuilt:
>
> [...]
>
> b. If the target *does* need to be rebuilt (is out-of-date), the
> pathname found during directory search is *thrown away*, and
> the target is rebuilt using the file name specified in the
> makefile. In short, if `make' must rebuild, then the target
> is rebuilt locally, not in the directory found via directory
> search.
>
> So, what happens? Well, make wants to build bar.new. So it looks for
> a bar.old, and sees it in directory foo via VPATH. Then it wants to
> build foo/bar.old, and finds an RCS file that's newer in foo/RCS.
>
> Since the prerequisite is newer, the target needs to be rebuilt, so by
> rule (4b) above the VPATH pathname is thrown out and make runs the rule
> to build just "bar.old". The rule you quote for RCS says that files
> that exist won't be updated, but "bar.old" _doesn't_ exist (foo/bar.old
> does, but that's not what we're building), so it _is_ updated.
>
> This seems to me to be correct and documented behavior.
>
> I _think_ what you want to do is add a "GPATH := foo" to your makefile.
> That will tell make that "foo" is a "global directory" and that make
> shouldn't throw out pathnames when targets are found in that directory.
> See the make manual docs for GPATH.
>
