By posting here I am hoping to attract the attention of people who will
be attending the Perl Toolchain Summit
(https://perltoolchainsummit.org/pts2024/) to be held in Lisbon from
April 25 to 28. (I originally posted this message to perl5-porters on
March 28.) We are experiencing problems with infrastructure that plays
a vital role in our ongoing work on the Perl core distribution --
problems which may benefit from the attention of PTS attendees.
I. CPANtesters not reporting names of distributions tested
I originally reported this problem on the cpan-testers-discuss mailing
list/newsgroup on January 9 2024
(https://www.nntp.perl.org/group/perl.cpan.testers.discuss/2024/01/msg4599.html).
New versions of certain distributions have been uploaded to CPAN and
have been tested by various CPANtesters rigs -- but the names of those
distributions are not being logged and, as an apparent consequence, no
reports for those versions are being presented in the web interfaces.
Case in point: In my original posting I reported that Net-SSLeay-1.94
was uploaded on January 8 (which was off by one day; the upload was on
January 7) but no page displaying results for that version had yet to
appear either at the matrix
(http://matrix.cpantesters.org/?dist=Net-SSLeay) or the fast-matrix
(http://fast-matrix.cpantesters.org/?dist=Net-SSLeay). On both pages the
*displayed* version is 1.93_05, but you can also see:
#####
Other versions
NOTE: no report for latest version 1.94
#####
So the system is aware that 1.94 has been released to CPAN but believes
that no reports have yet been uploaded. That means that we at Perl 5
Porters have no way of knowing whether changes we have been making in
the core distribution may have broken Net-SSLeay -- a distribution for
which there has even been discussion of shipping it with core.
This problem persists. Another CPAN distribution whose latest version
is not being reported is MIME-tools-5.514, uploaded to CPAN on February
6. The most recent version reported on at the matrix or the fast-matrix
(http://fast-matrix.cpantesters.org/?dist=MIME-tools) is 5.513, the last
report for which was received on February 6.
Slaven Rezić reported this problem as an issue in the
cpantesters-backend bug queue on February 21
(https://github.com/cpan-testers/cpantesters-backend/issues/27). He
suggested a possible diagnosis and correction. While there has been
some discussion in that ticket, it does not appear to have attracted the
attention of people who to my knowledge were involved with the
CPANtesters infrastructure in the past.
Implication: Blead may be "breaking" CPAN, but, less than a month going
into our next production release, we are not finding out about it.
II. Delayed mail forwarding through cpan.org
On February 10 I reported to P5P
(https://www.nntp.perl.org/group/perl.perl5.porters/2024/02/msg267842.html)
that I was experiencing delays in receiving email notifications about
postings being made to our GitHub issues and pull request queues. At
the time I speculated that the problem might have been on GH's side.
This was incorrect. Additional data and discussion with the perl.org
system administrators led the admins to post
(https://log.perl.org/2024/02/cpanorg-email-deliverability-issues.html)
this was affecting all cpan.org email addresses, was a result of some
backend changes and that "addressing this issue will require significant
technical adjustments."
My hunch is that these problems are as much human (insufficient
volunteer resources; recruiting the next generation of people to
maintain Perl's infrastructure) as they are technical. I hope they can
be placed on the agenda for the PTS.
Thank you very much.
Jim Keenan
