On Sat, Dec 18, 2004 at 07:13:59AM -0500, Randy W. Sims wrote: > dmake test TEST_VERBOSE=1 > dmake.exe: makefile: line 807: Warning -- Macro `TEST_VERBOSE' cannot > be redefined > ...
Sure, but did that actually produce verbose test output? If it did then its warning that the Makefile is trying to redefine the command line macro making that one of the most pointless warnings ever. > Manually changing the generated F<Makefile> to use the C<MACRO *= LINE> > form of assignment fixes the problem. But that seems like a PITA to have > to specify this alternative syntax for all possible commandline macros. Science does not have numbers small enough to express the odds of that happening. > How is MakeMaker different now than before? I scanned the dmake docs, > hoping to find some type of directive to change the behavior, but I > don't see anything. Before... how before? What's the newest MakeMaker where you don't see this warning? > Further, note that the latest trunk tarball fails F<basic.t> test #74, > but only with dmake. As before, all tests pass with nmake. Note that > this is the same perl; I just changed Config.pm to switch dmake/nmake. > > ok 73 - init_dirscan skipped distdir > # Failed test (t/basic.t at line 271) > # got: '65280' > # expected: '0' > # C:\devel\perl\bin\perl.exe -e "chdir 'Liar'; system > 'x:\opt\dmake\dmake.exe -S clean' if -f 'Makefile';" That's "make realclean" calling the clean target on a subdir. Any idea what its complaining out there? That's just: $(PERLRUN) -e "chdir 'Liar'; system '$(MAKE) clean' if -f $(FIRSTMAKEFILE) which is the cross platform version of cd Liar && -f $(FIRSTMAKEFILE) && $(MAKE) clean You can see the code for that in clean_subdirs_target() now moved to MM_Any. You might want to make sure you're up to date and didn't get nailed by some stupid mistake of mine from tonight. -- Michael G Schwern [EMAIL PROTECTED] http://www.pobox.com/~schwern/ Don't worry, baby, my wrath can be pretty groovy. http://www.goats.com/archive/980804.html
