| Subject: | 0001-Makefile-should-depend-upon-VERSION_FROM.patch |
When make testing new pulls of bleadperl, I typically just do
make test, skipping the git clean -dxf or similar.
When a recent commit has bumped a module's VERSION,
I get errors during test, like:
cpan/Unicode-Normalize/t/tie................................... \
Unicode::Normalize object version 1.10 does not match \
bootstrap parameter 1.12 at ../../lib/DynaLoader.pm line 213.
I usually just find and delete the Makefile, and let
Makefile.PL && ExtUtils-MakeMaker rebuild it.
Typically tests will then pass.
to-be-attached patch fixes this by patching MM_Unix.pm to
add $(VERSION_FROM) dependency to $(FIRST_MAKEFILE)
Ive kludged t/pm_to_blib.t to run make 2x after touching Dummy.pm,
1st make errors out (which is intended, perhaps to compensate
for a lack of some dependencies), 2nd succeeds, and returns its
exit code.
Note: this doesnt fix all variety of errors in incremental makes,
but does address a prevalent mode of failure.
tested on [jimc@groucho ExtUtils-MakeMaker]$ git describe
v6.57_11-2-g49dd39f
and on perl [jimc@groucho perl]$ git describe
v5.14.0-618-gec39a4a
which is a merge on top of
ecb7327 perldelta for breakpoints by file name
| Subject: | 0001-Makefile-should-depend-upon-VERSION_FROM.patch |
From 49dd39fa65ebd0105d937275fba7dcd2871077ff Mon Sep 17 00:00:00 2001
From: Jim Cromie <jim.cromie@gmail.com>
Date: Thu, 16 Jun 2011 13:21:47 -0600
Subject: [PATCH] Makefile should depend upon $VERSION_FROM
RFC: has minor kludge in test file.
MM_Unix.pm: add $(VERSION_FROM) to $(FIRST_MAKEFILE) dependencies.
This cause regeneration of Makefile when the VERSION is updated in
the source-file.
Going forward, the dependency fixes problems like following,
when perl source is pulled, and incremental make test is done:
cpan/Unicode-Normalize/t/tie................................... \
Unicode::Normalize object version 1.10 does not match \
bootstrap parameter 1.12 at ../../lib/DynaLoader.pm line 213.
Also patches t/pm_to_blib.t, to run make 2x after lib/Big/Dummy.pm
is updated. 1st make regenerates Makefile (as intended), but it returns
failure (also intended, unclear why). 2nd make covers that failure
by returning its success, so test passes.
NOTE: "$make 2>/dev/null ; $make" is not portable,
if the concept has merit, a portable way can be found.
---
lib/ExtUtils/MM_Unix.pm | 2 +-
t/pm_to_blib.t | 2 +-
2 files changed, 2 insertions(+), 2 deletions(-)
diff --git a/lib/ExtUtils/MM_Unix.pm b/lib/ExtUtils/MM_Unix.pm
index a3b7e9d..082a962 100644
--- a/lib/ExtUtils/MM_Unix.pm
+++ b/lib/ExtUtils/MM_Unix.pm
@@ -2524,7 +2524,7 @@ $(OBJECT) : $(FIRST_MAKEFILE)
$m .= sprintf <<'MAKE_FRAG', $newer_than_target, $mpl_args;
# We take a very conservative approach here, but it's worth it.
# We move Makefile to Makefile.old here to avoid gnu make looping.
-$(FIRST_MAKEFILE) : Makefile.PL $(CONFIGDEP)
+$(FIRST_MAKEFILE) : Makefile.PL $(CONFIGDEP) $(VERSION_FROM)
$(NOECHO) $(ECHO) "Makefile out-of-date with respect to %s"
$(NOECHO) $(ECHO) "Cleaning current config before rebuilding Makefile..."
-$(NOECHO) $(RM_F) $(MAKEFILE_OLD)
diff --git a/t/pm_to_blib.t b/t/pm_to_blib.t
index 66ef879..8035af6 100644
--- a/t/pm_to_blib.t
+++ b/t/pm_to_blib.t
@@ -54,7 +54,7 @@ my $make = make_run();
print $fh "Something else\n";
close $fh;
- run_ok($make);
+ run_ok("$make 2>/dev/null;$make");
like slurp("blib/lib/Big/Dummy.pm"), qr/Something else\n$/;
}
--
1.7.4.4