develooper Front page | perl.perl5.porters | Postings from April 2018

CPAN-river-1000: failures as of perl-5.27.11

Thread Next
From:
James E Keenan
Date:
April 21, 2018 15:39
Subject:
CPAN-river-1000: failures as of perl-5.27.11
Message ID:
20180421153945.4879.qmail@lists-nntp.develooper.com
You can find data from a test run of the "CPAN-river-top-1000" against
perl-5.27.11 here:

http://thenceforward.net/perl/misc/cpan-river-1000-perl-5.27-master.psv.gz

The file above presents the data in the same format which I used for
5.27.6 and 5.27.7 in November and December 2017.  What is essentially
the same data is presented in a slightly different format here:

http://thenceforward.net/perl/misc/xformat-cpan-river-1000-perl-CPAN-river-1000: 
failures as of perl-5.27.105.27-master.psv.gz

The attachment, 'changes-5.27.10-to-5.27.11.csv', presents another
subset of the data set found at the second link above.  It presents
results from just the last two monthly releases for those distributions
which either (a) had a change of grade between 5.27.10 and 5.27.11; or
(b) received a grade other than 'PASS' in 5.27.11.

A large number of the distributions listed in
changes-5.27.10-to-5.27.11.csv have an 'x' in the final column, meaning
that when I ran my test-against-dev program earlier this morning,
'cpanm' bailed out before being able to test that distribution.  That
generally happens when an upstream prerequisite has a build or test
failure, fails to install, and causes 'cpanm' to bail when it comes to
the prerequisite's reverse dependency.  (If I were using 'cpan' rather
than 'cpanm', this would be logged as a DISCARD in the
.cpanreporter/reports-sent.db file; thanks to Slaven for calling my
attention to that.)

This month, many of those 'x' distributions were affected by problems
with File-Copy-Recursive.  Version 0.41 of that distribution was
released two days ago, but failed to specify one dependency.  The author
has subsequently corrected that omission and has issued version 0.42 --
but that didn't take place until after my test-against-dev program had
run.  Many distributions in the CPAN river top 1000 are dependent on
File-Copy-Recursive, so they were not reached when test-against-dev ran.
Now that FCR 0.42 has been released, I'm going to retry the
distributions that were graded 'x'.

Some of the distributions in changes-5.27.10-to-5.27.11.csv are
previously known failures.  Dave Mitchell surveyed these on this list
last month:
https://www.nntp.perl.org/group/perl.perl5.porters/2018/03/msg250135.html

Anyone who wishes to follow up on this data or help me further develop
analyses like this should contact me.  This will be the final report of
this data in the 5.27.* development cycle.  I hope to begin anew,
covering a larger subset of the CPAN river, once 5.29.0 is released.

Thank you very much.
Jim Keenan

Thread Next


nntp.perl.org: Perl Programming lists via nntp and http.
Comments to Ask Bjørn Hansen at ask@perl.org | Group listing | About