Re: Failure in test subobj9.test with heirloom make

[Ralf Wildenhues]

Hi Stefano,

* Stefano Lattarini wrote on Tue, Apr 20, 2010 at 04:50:54PM CEST:
> While trying out the Automake testsuite using Heirloom make as $MAKE. 
> I encounterd a failure in test `subobj9.test'.  Attached here are the 
> test log and the testscript-generated Makefile and Makefile.in heirloom 
> make is chocking upon.  Let me know if you need more information.

> + /opt/heirloom/bin/make print
> make: fatal error: line 273: cannot get /src/.deps/bar.Plo for including
> .

Interesting.  Does the patch below make the test pass for this make, or
does only it fail later on then?  If it passes, does it also suffice to
only add the first of the two added lines?

I'm trying to understand which of these lines this make interprets
wrongly:
  include file
  include ./file
  include .//file
  include sub/file
  include sub//file
  include ./sub/file
  include .//sub//file
  include "file"
  include "./file"
  include ".//file"
  include "sub/file"
  include "sub//file"
  include "./sub/file"
  include ".//sub//file"

because we might just be able to improve the test in m4/make.m4 to
detect that limitation.

Thanks,
Ralf

diff --git a/automake.in b/automake.in
index f25674b..d34e1dd 100644
--- a/automake.in
+++ b/automake.in
@@ -2142,6 +2142,8 @@ sub handle_single_transform ($$$$$%)
            my $depfile = $object;
            $depfile =~ s/\.([^.]*)$/.P$1/;
            $depfile =~ s/\$\(OBJEXT\)$/o/;
+           $depfile =~ s,//*,/,;
+           $depfile =~ s,^\./,,;
            $dep_files{dirname ($depfile) . '/$(DEPDIR)/'
                         . basename ($depfile)} = 1;
        }