> On Aug 3, 2006, at 10:11 PM, Ken Williams wrote: > I think I spoke too soon. I get about a bazillion errors. However, > I also get these errors *without* the patch. Do you know, is there a > new minimum version of CPANPLUS I need now? I've got version 0.055 > installed.
0.072 is most recent, 0.73_01 is on CPAN as a dev release. Those are
best to test against, but there should be no API changes or things like
that to bite you.
> I'm running this all using the current SVN version from http://
> svn.perl.org/modules/CPANPLUS-Dist-Build/trunk/ .
Ouch.
I've looked into this, by checking out a copy from svn.perl.org and trying
to 'fix' what's wrong. It seems to be a combination of factors :(
> t/02_CPANPLUS-Dist-Build..............ok 4/0Couldn't read chunk at
> offset unknown at /Library/Perl/5.8.1/CPANPLUS/Internals/Extract.pm
> line 158
The test packages in t/src/* seem to be corrupt :(
$ gzip -d Foo-Bar-0.01.tar.gz
gzip: Foo-Bar-0.01.tar.gz: invalid compressed data--crc error
Attached are the 2 test packages from my p4 repository. I copied them
over to their respective locations in the svk checkout, and all went well.
I can only assume somehow they got mangled... Unfortunaly, svk diff
doesn't diff binary files, so i had to attach them.
Please copy Foo-Bar-0.01.tar.gz.NOXS to t/src/noxs/Foo-Bar-0.01.tar.gz
Please copy Foo-Bar-0.01.tar.gz.XS to t/src/xs/Foo-Bar-0.01.tar.gz
After fixing this, I noticed that the c-d-build cpan-testers patch i sent was
the wrong one (shame on me :( ) -- i sent you the one for c-d-MM instead.
Unfortunately, it applied 'cleanly' as those sections of code are similar in
both files... it did however break all the tests :(
I've attached a patch, against your svn repository, that holds the *right*
patch set this time, which should do the trick;
$ make test PERL_DL_NONLAZY=1 /opt/bin/perl "-MExtUtils::Command::MM"
"-e" "test_harness(0, 'blib/lib', 'blib/arch')" t/*.t
t/01_CPANPLUS-Dist-Build-Constants....ok
t/02_CPANPLUS-Dist-Build..............ok
All tests successful.
Files=2, Tests=37, 6 wallclock secs ( 2.70 cusr + 2.39 csys = 5.09 CPU)
Now, I'm seeing the same thing happening to you as was happening to me:
I was too scared of the M::B code to actually feel confident about making
changes to C::D::Build any longer, afraid of breaking things.
It seems you feel the same way about CPANPLUS (and probably rightfully so).
Perhaps only 1 of the 2 of us maintaining this module is a recipe for
trouble.
So, for your consideration, i suggest a commit bit for myself to apply
CPANPLUS
related patches to C::D::Build in your repository. The module-build crew
would still be the ones releasing this code, as it ties in much more
tightly with
M::B than cpanplus nowadays.
Of course, I'm also happy to keep an eye on the c-d-build queue in RT when
CPANPLUS specific issues pop up and apply patches/fixes as needed. For this,
you'd need to list me as co-maintainer again in PAUSE.
This is merely a suggestion, in the hope we can ship a better, allround
product.
I won't be offended if you dismiss it, but i thought it best to offer :)
-- Jos
c-d-build-comprehensive.patch
Description: Binary data
Foo-Bar-0.01.tar.gz.NOXS
Description: Binary data
Foo-Bar-0.01.tar.gz.XS
Description: Binary data
