outstanding RT issues

[karenetheridge]

These aren't linked via the resources bugtracker (that link goes to the main perl5 queue, which is correct since this dist is core-first).

However, there are issues outstanding at https://rt.cpan.org/Dist/Display.html?Queue=ExtUtils-Manifest

[karenetheridge]

When all these tickets are resolved, we should probably close the queue to avoid future misunderstandings.

[karenetheridge]

They are:

• https://rt.cpan.org/Ticket/Display.html?id=21283 ExtUtils::Manifest does not follow symlinks
• https://rt.cpan.org/Ticket/Display.html?id=57046 Tests fail if $ENV{PERL_MM_MANIFEST_VERBOSE} is set to a false value.
• https://rt.cpan.org/Ticket/Display.html?id=3109 Aborting 'make manifest' can leave you with a broken MANIFEST
• https://rt.cpan.org/Ticket/Display.html?id=2746 maniread() should return a tied, case-insensitive hash on VMS
• https://rt.cpan.org/Ticket/Display.html?id=21290 print to STDERR instead of warn in ExtUtils::Manifest
• https://rt.cpan.org/Ticket/Display.html?id=57507 Default MANIFEST.SKIP too aggressive about exclusions
• https://rt.cpan.org/Ticket/Display.html?id=57586 whole-name-quoted filename can't be read

[karenetheridge]

And the tickets in the perl5 queue, courtesy of Jim Keenan:

Here are the
rt.perl.org tickets and my action recommendations for each.

https://rt.perl.org/Ticket/Display.html?id=122415
whole name quoted filename
Review the patch I submitted:
https://rt.perl.org/Ticket/Attachment/1302699/690549/122415-0001-Have-maniread-properly-handle-whole-name-quoted-file.patch

https://rt.perl.org/Ticket/Display.html?id=122416
MANIFEST.SKIP exclusions
Note that OP has responded. Maintainer should make a decision as to
whether change is desired and work it out with OP.

https://rt.perl.org/Ticket/Display.html?id=122421
aborting make manifest
Maintainer (Toolchain Gang, I guess) should make a decision as to
whether change is desired

https://rt.perl.org/Ticket/Display.html?id=122417
print STDERR vs warn
Maintainer should make a decision as to whether change is desired.
Patch will be simple.

https://rt.perl.org/Ticket/Display.html?id=122420
maniread on VMS
Make friends with Craig Berry :-)

[jkeenan]

Toolchainers:

I direct your attention to https://rt.perl.org/Ticket/Display.html?id=122415#txn-1303551, in which Ricardo Signes states that the patch I submitted in that ticket to correct a problem in ExtUtils-Manifest addresses his original complaint (first filed in https://rt.cpan.org/Ticket/Display.html?id=57586).

At this point, if everyone were in complete agreement as to "where ExtUtils-Manifest is being maintained," I would apply the patch to Perl 5 blead and close RT 122415.

However, (a) I have been told that ExtUtils-Manifest is being maintained by the Toolchain Gang; and (b) the evidence in Porting/Maintainers.pl with respect to this distribution is ambiguous. The entry for EU-M in that file is:

    'ExtUtils::Manifest' => {
        'DISTRIBUTION' => 'BINGOS/ExtUtils-Manifest-1.64.tar.gz',
        'FILES'        => q[dist/ExtUtils-Manifest],
        'EXCLUDED'     => [qr(^xt/)],
    },

On the one hand, the fact that the top-level directory in the value for 'FILES' is 'dist' suggests that the library is maintained "in blead", i.e., by Perl 5 Porters. From inline comments in Maintainers.pl:

# UPSTREAM indicates where patches should go.  This is generally now
# inferred from the FILES: modules with files in dist/, ext/ and lib/
# are understood to have UPSTREAM 'blead', meaning that the copy of the
# module in the blead sources is to be considered canonical, while
# modules with files in cpan/ are understood to have UPSTREAM 'cpan',
# meaning that the module on CPAN is to be patched first.

On the other hand, the fact that the value of 'DISTRIBUTION' begins with 'BINGOS' suggests that EU-M ought to be thought of as being maintained "on CPAN", i.e., not by Perl 5 Porters in blead. From other inline comments in Maintainers.pl:

# DISTRIBUTION names the tarball on CPAN which (allegedly) the files
# included in core are derived from. Note that the file's version may not
# necessarily match the newest version on CPAN.

Note that if ExtUtils-Manifest is primarily maintained "on CPAN", then inside the Perl 5 core distribution it ought to be under the 'cpan/' directory, not under the 'dist/' directory (as is the case, e.g., for ExtUtils-MakeMaker).

I would like to ask that the members of the Toolchain Gang discuss this issue -- preferably in collaboration with Ricardo -- and work out a decision as to where ExtUtils-Manifest is to reside. I think these are the possible scenarios:

• If Toolchain Gang takes maintenance, then:
  • The rt.cpan.org bug queue should be closed to new entries and the preferred bug tracker should be moved to https://github.com/Perl-Toolchain-Gang/ExtUtils-Manifest/issues.
  • Language inside the distribution should be changed as needed so as to correctly identify who's maintaining the distro, where to file bugs, etc.
  • Inside the Perl 5 core distribution, EU-M moves from dist/ to cpan/.
  • The Toolchain Gang should apply the patch in RT 122415 in github, release a new version to CPAN, then have the version in Perl 5 blead brought up to date with the new CPAN version.
• If Perl 5 Porters takes maintenance, then:
  • The rt.cpan.org bug queue should be closed to new entries. (That queue already indicates that bugs should be filed via mail to [email protected].)
  • The existing CPAN distribution of EU-M should be closed down.
  • Inside the Perl 5 core distribution, EU-M should move from dist/ to lib/.
  • The entry in Porting/Maintainers.pl should have the 'DISTRIBUTION' key removed, as there will no longer be a distribution found on CPAN.
  • I will apply the patch in RT 122415 directly to blead and close that ticket.

I think of myself only as a peripheral member of the Toolchain Gang, so I will respect whatever decision you and the pumpking work out.

(But I do like to apply approved patches quickly, particularly when they're my patches! :-) )

Thank you very much.
Jim Keenan

[dagolden]

I vote for PTG taking maintenance and following Jim's action steps, except that I don't see any reason to "close" the rt.cpan.org queue. The split reporting of tickets here and on RT is a well known "feature" of RT and I think we'll need to find better ways to coordinate across. I'm fine having the bug-reporting metadata point to Github if people so desire.

[karenetheridge]

On Tue, Aug 05, 2014 at 04:50:22PM -0700, David Golden wrote:

I vote for PTG taking maintenance and following Jim's action steps, except that I don't see any reason to "close" the rt.cpan.org queue. The split reporting of tickets here and on RT is a well known "feature" of RT and I think we'll need to find better ways to coordinate across. I'm fine having the bug-reporting metadata point to Github if people so desire.

Note that I requested that the RT.cpan queue be closed with the
understanding that ExtUtils-Manifest would remain a blead-first dual-life
module. I wasn't aware of BinGOs' intention to switch it to cpan-first.

In that case, please please don't close any RT queues (unless you're the
sole maintainer of a dist - then it's your own decision) -- filing a github
issue requires creating an account, which adds (IMHO) unreasonable and
unnecessary barriers.

I'm also totally fine with the dist shifting to cpan-first. It will
probably make the maintenance a bit easier, as PTG members can then freely
merge to the github distribution and not worry as much about getting
patches into blead (but I'm willing to keep feeding those into perlbug, as
I started doing last month for a few old PRs that needed some attention).

[dagolden]

I think having it upstream CPAN will make it easier to iterate it as necessary.

blead upstream is best, IMO, for things that are stable, under-maintained or very platform specific. E.g. I lobbied for EU::CBuilder to go upstream blead because it wasn't very maintained and compiler expertise was more likely to be found in p5p than elsewhere.

[bingos]

I never said nor implied it should be upstream CPAN. Other 'maintained in core' distributions have DISTRIBUTION set in Porting/Maintainers.pl too, it doesn't imply they are CPAN upstream either.

The very fact that we are having this conversation would suggest that the maintenance burden should shift to PTG and upstream be CPAN.

[jkeenan]

David, Karen, Chris,

Thanks for your quick response. If you/we are in consensus that ExtUtils-Manifest should move to "upstream CPAN", can you file a perlbug to that effect? I will then carry out the second and third second-level bullet points under the "Toolchain Gang takes maintenance" bullet point in my earlier post.

Also let me know whether you want that patch applied to EU-M in blead first (which would immediately close perl:122415.

I leave the question of where EU-M's bugs should be tracked to you, as in that case it will no longer be rt.perl.org.

[Leont]

I agree with the others, upstream CPAN is generally preferable unless there is a good reason otherwise.

[dagolden]

@jkeenan → https://rt.perl.org/Ticket/Display.html?id=122483

[jkeenan]

 Have taken that ticket.  Will handle by weekend.  On 08/06/14, David [email protected] wrote: @jkeenan → https://rt.perl.org/Ticket/Display.html?id=122483—Reply to this email directly or view it on GitHub.

[jkeenan]

Do you have a decision on https://rt.perl.org/Ticket/Display.html?id=122415, i.e., shall I apply the patch to p5 blead?

Correction: I see that you want to apply the patches to github.

Thank you very much.
Jim Keenan

[jkeenan]

Can I ask where things stand with respect to https://rt.perl.org/Ticket/Attachment/1302699/690549/122415-0001-Have-maniread-properly-handle-whole-name-quoted-file.patch ?

Do I need to open a separate github issue for someone in the Gang to apply it, release a new version to CPAN, then prepare a patch for Perl 5 blead?

I would like to be able to close out https://rt.perl.org/Ticket/Display.html?id=122415. (Actually, I'd like to be able to close out all the EU-M related tickets at rt.perl.org, but not until you've made the changes upstream first.)

Thank you very much.
Jim Keenan

[bingos]

I've released 1.66 which resolves the following

• https://rt.cpan.org/Ticket/Display.html?id=57586 whole-name-quoted filename can't be read
• https://rt.cpan.org/Ticket/Display.html?id=21283 ExtUtils::Manifest does not follow symlinks
• https://rt.cpan.org/Ticket/Display.html?id=57046 Tests fail if $ENV{PERL_MM_MANIFEST_VERBOSE} is set to a false value.

[jkeenan]

Chris, thanks for looking into this. Will you be synching Perl 5 core with this?

Thank you very much.
Jim Keenan

[bingos]

It was synchronised almost immediately:
http://perl5.git.perl.org/perl.git/commitdiff/9a7aaf5783774d0974f8e80c4d6572d213f7fd92

[mohawk2]

Is this still a live issue, or should it be closed?

[karenetheridge]

Is this still a live issue?

Yes, there are lots of open issues tracked here. They could perhaps be opened as separate issues, but until then, this should stay open.

[dolmen]

I just closed RT#21283, RT#57586 and RT#57046 as they have been fixed since 1.66. @bingos++

[karenetheridge]

@dolmen++