Closed GlennWood closed 2 years ago
From the
Makefile.PL
step, we get:Can't exec "cc": No such file or directory at //extsrc/perl/5.32.1d/rs6000_71/lib/5.32.1/ExtUtils/CBuilder/Base.pm line 342. No compiler found, won't generate 'script/parl!
Can you attach the full transcript (stdout and stderr) of a clean build of PAR::Packer, i.e.
your/perl Makefile.PL
make
make test
In case you applied any patches after the unpack step, please post them as well.
attached log files withCC.log
and withoutCC.log
are the result of:
bash-4.4$ ( /extsrc/perl/5.32.1e/rs6000_71/bin/perl Makefile.PL INSTALL_BASE=$INSTALL_BASE CC=$CC && make manifest && make && make install && make dist ) >../withCC.log 2>&1
bash-4.4$ make clean
bash-4.4$ ( /extsrc/perl/5.32.1e/rs6000_71/bin/perl Makefile.PL INSTALL_BASE=$INSTALL_BASE && make manifest && make && make install && make dist ) >../withoutCC.log 2>&1
Note that the make clean
has a flaw:
rm -f 110 110OBJECTS)
/bin/sh: syntax error at line 1 : `)' unexpected
Note that the
make clean
has a flaw:rm -f 110 110OBJECTS) /bin/sh: syntax error at line 1 : `)' unexpected
One line fix:
diff --git a/myldr/Makefile.PL b/myldr/Makefile.PL
index 60458d5..c2d4559 100644
--- a/myldr/Makefile.PL
+++ b/myldr/Makefile.PL
@@ -256,7 +256,7 @@ $res_section
clean::
-\$(RM_F) boot_embedded_files.c my_par_pl.c
- -\$(RM_F) $(OBJECTS)
+ -\$(RM_F) \$(OBJECTS)
-\$(RM_F) *.opt *.pdb perlxsi.c
-\$(RM_F) usernamefrompwuid.h
-\$(RM_F) $par_exe $boot_exe @parl_exes Dynamic.pm Static.pm
Thanks for the logs. Why do you make manifest
and make dist
? These targets are for me (to generate the release tarball for uploading to CPAN), but of no use for somone just building and installing PAR::Packer. But aside from copious output from xlc they look OK.
make manifest
and make dist
because we maintain our own cpansite with fully vetted packages.
I have incorporated a configured CC path into our build scripts and this is working fine now. You can close this "informational" issue now. Thank you, Roderich!
make manifest and make dist because we maintain our own cpansite with fully vetted packages.
I see.
Thanks for following up.
While hacking on #62, a message flashed on my console that seems intriguing. This information might be enlightening regarding issue #64.
From the
Makefile.PL
step, we get:It spuns by so quickly that I only recently noticed it. And since the build proceeded all the way through, I figured it doesn't matter.
However, this is especially intriguing because it was a change in the CC compiler made by our DevOps that eventually led up to our conundrums with rs6000. DevOps changed the CC from
cc
(found on the PATH) to/opt/IBM/xlc/16.1.0/bin/xlc
(not on the PATH, neither iscc
on their new build-host), and we've had issues with it ever since.In my hack on
fix-pp-clean
branch, I did addCC=/opt/IBM/xlc/16.1.0/bin/xlc
to the Makefile.PL step. This message went away (but unfortunately the "Not a typewriter" remained).I am going to incorporate the
CC=
into theCPANConfig.pm[makepl_arg]
property and see what happens. Again, I expect the message to go away, but the failure to remain, but this might be enlightening vis-a-vis #64.