Re: [PATCH 4/4] Move initialization of Git::SVN variables into Git::SVN.

2 messages, 2 authors, 2016-06-15 · open the first message on its own page

Re: [PATCH 4/4] Move initialization of Git::SVN variables into Git::SVN.

From: Junio C Hamano <hidden>
Date: 2016-06-15 22:54:21

Eric Wong [off-list ref] writes:
quoted hunk
OK, all fixed, all I needed was this (squashed in):
--- a/perl/Makefile
+++ b/perl/Makefile
@@ -22,6 +22,8 @@ clean:
 	$(RM) $(makfile).old
 	$(RM) PM.stamp
 
+$(makfile): PM.stamp
+
 ifdef NO_PERL_MAKEMAKER
 instdir_SQ = $(subst ','\'',$(prefix)/lib)

-------------
The redundant dependencies are biting us :<  I agree there presence in
the top-level Makefile needs to be reviewed.
Do you feel confident enough that we can leave that question hanging
around and still merge this before 1.7.12 safely?

I do not think it is a regression at the Makefile level per-se---we
didn't have right dependencies to keep perl.mak up to date, which
was the root cause of what we observed.

But the lack of dependencies did not matter before this series
because the list of *.pm files never changed, so in that sense the
series is what introduced the build regression, and I do not have a
solid feeling that we squashed it.
Anyways, you can pull this now from my master:

The following changes since commit cdd159b2f56c9e69e37bbb8f5af301abd93e5407:

  Merge branch 'jc/test-lib-source-build-options-early' (2012-07-25 15:47:08 -0700)

are available in the git repository at:

  git://bogomips.org/git-svn master

for you to fetch changes up to 5c71028fced46d03bf81b8625680d9ac87c8f4f0:

  Move initialization of Git::SVN variables into Git::SVN. (2012-07-27 22:14:54 +0000)

----------------------------------------------------------------
Junio C Hamano (1):
      perl: detect new files in MakeMaker builds

Michael G. Schwern (7):
      Quiet warning if Makefile.PL is run with -w and no --localedir
      Don't lose Error.pm if $@ gets clobbered.
      The Makefile.PL will now find .pm files itself.
      Extract some utilities from git-svn to allow extracting Git::SVN.
      Prepare Git::SVN for extraction into its own file.
      Extract Git::SVN from git-svn into its own .pm file.
      Move initialization of Git::SVN variables into Git::SVN.

 Makefile                       |    7 +
 git-svn.perl                   | 2340 +---------------------------------------
 perl/.gitignore                |    1 +
 perl/Git/SVN.pm                | 2324 +++++++++++++++++++++++++++++++++++++++
 perl/Git/SVN/Utils.pm          |   59 +
 perl/Makefile                  |    5 +
 perl/Makefile.PL               |   35 +-
 t/Git-SVN/00compile.t          |    9 +
 t/Git-SVN/Utils/can_compress.t |   11 +
 t/Git-SVN/Utils/fatal.t        |   34 +
 10 files changed, 2487 insertions(+), 2338 deletions(-)
 create mode 100644 perl/Git/SVN.pm
 create mode 100644 perl/Git/SVN/Utils.pm
 create mode 100644 t/Git-SVN/00compile.t
 create mode 100644 t/Git-SVN/Utils/can_compress.t
 create mode 100644 t/Git-SVN/Utils/fatal.t

Re: [PATCH 4/4] Move initialization of Git::SVN variables into Git::SVN.

From: Eric Wong <hidden>
Date: 2016-06-15 22:54:21

Junio C Hamano [off-list ref] wrote:
Eric Wong [off-list ref] writes:
quoted
The redundant dependencies are biting us :<  I agree there presence in
the top-level Makefile needs to be reviewed.
Do you feel confident enough that we can leave that question hanging
around and still merge this before 1.7.12 safely?
Yes.
I do not think it is a regression at the Makefile level per-se---we
didn't have right dependencies to keep perl.mak up to date, which
was the root cause of what we observed.

But the lack of dependencies did not matter before this series
because the list of *.pm files never changed, so in that sense the
series is what introduced the build regression, and I do not have a
solid feeling that we squashed it.
Right, I agree the original dependencies are not good and it's not
a recent regression in the Makefile level.

I do feel our patch deals with the problem for now.  I've been going
between commits in Michael's 3rd series and haven't noticed new issues
when running the tests.
Keyboard shortcuts
hback out one level
jnext message in thread
kprevious message in thread
ldrill in
Escclose help / fold thread tree
?toggle this help