ack requires two attempts to install on Perl 5.20.0

[petdance]

From @rjbs on Twitter "The only problem I had migrating laptop to 5.20.0 was that I had to make two attempts to install ack. 2nd attempt worked. Known?"

See https://gist.github.com/rjbs/4d506feec4dbdde5cb36

[n1vux]

RJBS says OSX, cpanm.

This was reported on mailing list with older Perls, on OSX and Linode, with cpan or cpanm, in Feb, March.
https://groups.google.com/forum/#!msg/ack-users/QL-ENH4P0GY/P7K3D8UB838J File::Next being present but needing upgrade seemed to be consistent theme then.

My comment at the time was i'd seen similar issues with cpan/cpanp/cpanm needing upgraded version of a prereq that didn't involve Ack or File::Next. If it's a fault in Ack or File::Next build/make files, it's sadly not unique to ours.

[n1vux]

Keenan's from March 22 (continuing above thread ) attached build fail log

Happened again today while loading up a brand new laptop running Ubuntu 13.10. Attaching gzipped output log showing that, although App:Ack built, tested and installed successfully, 'standalone' failed dramatically. Second time I called 'cpanm App::Ack', everything DWIMmed.

https://groups.google.com/group/ack-users/attach/7ff301c50fb7b23f/app-ack-failure-build.log.gz?part=2&authuser=0&view=1

[karenetheridge]

One issue behind this is that File::Next often fails to install cleanly:

Fetching http://www.cpan.org/authors/id/P/PE/PETDANCE/File-Next-1.12.tar.gz
-> OK
Unpacking File-Next-1.12.tar.gz
Entering File-Next-1.12
Checking configure dependencies from META.json
Checking if you have ExtUtils::MakeMaker 0 ... Yes (6.98)
Configuring File-Next-1.12
Running Makefile.PL
Checking if your kit is complete...
Looks good
Generating a Unix-style Makefile
Writing Makefile for File::Next
Writing MYMETA.yml and MYMETA.json
-> OK
Checking dependencies from MYMETA.json ...
Checking if you have ExtUtils::MakeMaker 0 ... Yes (6.98)
Checking if you have Test::More 0.88 ... Yes (1.001002)
Checking if you have File::Spec 0 ... Yes (3.47)
Building and testing File-Next-1.12
cp Next.pm blib/lib/File/Next.pm
Manifying blib/man3/File::Next.0
PERL_DL_NONLAZY=1 /Users/ether/perl5/perlbrew/perls/20.0/bin/perl "-MExtUtils::Command::MM" "-MTest::Harness" "-e" "undef *Test::Harness::Switches; test_harness(0, 'blib/lib', 'blib/arch')" t/*.t
# Testing File::Next 1.12, Perl 5.020000, /Users/ether/perl5/perlbrew/perls/20.0/bin/perl
t/00-load.t ....... ok
t/api.t ........... ok
t/dirs.t .......... ok
t/basic.t ......... ok
t/dot.t ........... ok
t/methods.t ....... ok
t/from_file.t ..... ok
t/pod-coverage.t .. skipped: Test::Pod::Coverage 1.04 required for testing POD

#   Failed test 'NO_PARMS'
#   at t/everything.t line 92.
# +----+----------------------+----+----------------+
# | Elt|Got                   | Elt|Expected        |
# +----+----------------------+----+----------------+
# |   0|'t/swamp/0'           |   0|'t/swamp/0'     |
# |   1|'t/swamp/a/a1'        |   1|'t/swamp/a/a1'  |
# |   2|'t/swamp/a/a2'        |   2|'t/swamp/a/a2'  |
# |   3|'t/swamp/b'           |   3|'t/swamp/b'     |
# |   4|'t/swamp/b/b1'        |   4|'t/swamp/b/b1'  |
# |   5|'t/swamp/b/b2'        |   5|'t/swamp/b/b2'  |
# |   6|'t/swamp/c/c1'        |   6|'t/swamp/c/c1'  |
# |   7|'t/swamp/c/c2'        |   7|'t/swamp/c/c2'  |
# *   8|'t/swamp/linkdir/a1'  *    |                |
# *   9|'t/swamp/linkdir/a2'  *    |                |
# |  10|'t/zero.t'            |   8|'t/zero.t'      |
# +----+----------------------+----+----------------+
# Looks like you failed 1 test of 4.
t/everything.t .... 
Dubious, test returned 1 (wstat 256, 0x100)
Failed 1/4 subtests 
t/follow.t ........ ok
t/parms.t ......... ok
t/named-pipe.t .... ok
t/from_stdin.t .... ok
t/sort.t .......... ok
t/zero.t .......... ok
t/pod.t ........... ok

Test Summary Report
-------------------
t/everything.t  (Wstat: 256 Tests: 4 Failed: 1)
  Failed test:  4
  Non-zero exit status: 1
Files=16, Tests=83,  1 wallclock secs ( 0.08 usr  0.04 sys +  1.04 cusr  0.16 csys =  1.32 CPU)
Result: FAIL
Failed 1/16 test programs. 1/83 subtests failed.
make: *** [test_dynamic] Error 255
-> FAIL Installing File::Next failed. See /Users/ether/.cpanm/work/1401214279.31647/build.log for details. Retry with --force to force install it.
-> FAIL Installing the dependencies failed: Module 'File::Next' is not installed
-> FAIL Bailing out the installation for ack-2.13_06.

[karenetheridge]

Also: http://matrix.cpantesters.org/?dist=ack+2.12

[n1vux]

​​T he File::Next ​ prereq​ fail ​shown above ​ is related to ​ two File::Next issues​

• Make t/follow.t use a temp directory https://github.com/petdance/file-next/issues/17 and

• ​ Race condition causes test failure under parallel testing #16 ​ ​

  ​ https://github.com/petdance/file-next/issues/16 ​

(Kare n has good comment there already on #16. Kent's comment is that follow.t might not be the only parallel problem. )​

​WRTO the ACK crimson matrix, it is not mostly ​this issue as Cygwin and MSwin are mostly green on http://matrix.cpantesters.org/?dist=File-Next

​ ​ , as are Darwin and Linux mostly.

Also, it's unclear how this would install clean reliably the second time as reported on this bug, but parallel could work by accident N-1 times in N, which might appear as 2nd time the charm.

-- Bill

CONNECTING ​DETAILS if needed.​

​F rom that log by Karen Etheridge, ​ I highlight 3 lines

at t/everything.t line 92.

* 8|'t/swamp/linkdir/a1' * | |

* 9|'t/swamp/linkdir/a2' * | |

​ ​ t/swamp/linkdir/ should only exist during t/follow.t .

File::Next t/follow.t has good intention of cleaning up

END { unlink( keys %links ); }

Parallel testing will hose this (as would a prior fail in follow.t .)​

[hoelzro]

This has been fixed in the dev branch.