On 5/10/22 07:17, Alceu Rodrigues de Freitas Junior via cpan-testers-discuss wrote:
> Hello folks,
>
> I released yesterday (a little bit late, to be honest) newer versions of
> the CPAN Smoker for OpenBSD Vagrant boxes:
>
> https://app.vagrantup.com/arfreitas/boxes/openbsd-7.0-cpan-smoker-i386
>
> https://app.vagrantup.com/arfreitas/boxes/openbsd-7.0-cpan-smoker-amd64
>
> OpenBSD 7.1 is already available and I should release boxes for it as
> well in the next few days;
>


Alceu,

Let me describe my own OpenBSD needs and see if you can comment on them.

For the ongoing development of Perl itself ("Perl 5 Porters" or "P5P"), I do considerable smoke-testing on the *BSDs. While I'm primarily concerned with FreeBSD, I also do some smoke-testing of Perl 5 blead on OpenBSD.

I have available a FreeBSD-12 server provided by New York City BSD Users Group. On that server I have installed VMs for more advanced versions of FreeBSD, for OpenBSD and occasionally for NetBSD. I have gotten best results by going to vagrantup.com and downloading virtualbox versions of VMs presented under the "generic" name. Currently I have installed an OpenBSD-6.9 VM from which I generate smoke-test reports such as this: http://perl5.test-smoke.org/search?selected_osnm=openbsd&report_osnm=openbsd&selected_perl=all&selected_host=perl-reporter-11. In the normal course of events, I would soon be looking to upgrade to OpenBSD-7.1 by installing this VM: https://app.vagrantup.com/generic/boxes/openbsd7/versions/3.6.14.

Now, here's where things get complicated.

While smoke-testing of Perl 5 blead is my primary concern, there are also times when I need to install blead perl on my VMs for the purpose of investigating reports that changes in blead have begun to cause failures to build or test CPAN distributions. In P5P we refer to this as the "Blead Breaks CPAN" ("BBC") problem. BBC issues often require some kind of modification both in Perl 5 core and in the CPAN distribution in order to be deemed resolved. For thorough verification, we need to test on Perl 5 blead on a wide variety of platforms, including OpenBSD. And while we're at it, it makes sense to generate a CPANtesters report which provides public documentation that the BBC issue has been fixed. So that means that in my OpenBSD VM, I typically want to install Perl 5 blead, then install a CPANtesters stack against that blead perl, and then test the CPAN distribution in a way that generates a CPANtesters report.

Roughly speaking, that means this sequence of commands:

#####
$ sh ./Configure -des -Dusedevel -Uversiononly -Dprefix=${SOME_TEMPDIR} -Dman1dir=none -Dman3dir=none
$ make install
$ cd ${SOME_TEMPDIR}
# get and install 'cpanm'
$ ./bin/cpanm Task::CPAN::Reporter
$ ./bin/cpan -i Some::Module
#####

The problem I've been encountering recently is that the Task-CPAN-Reporter stack includes IO-Socket-SSL, and the CPAN maintainer of IO-Socket-SSL has given up on trying to maintain that module on OpenBSD. See, for example, https://github.com/noxxi/p5-io-socket-ssl/issues/115. While the OpenBSD people do a good job of maintaining their *port* of IO-Socket-SSL via patches, that port is only intended to work with the *system* perl in a given version of OpenBSD. If you try to use 'cpan' or 'cpanm' to install something that has IO::Socket::SSL in its dependency chain (e.g., Task::CPAN::Reporter), you will get test failures.

So I'm wondering how your OpenBSD VM, which is specifically intended to serve as the basis for CPANtesters reporting, work around this problem?

Thank you very much.
Jim Keenan
  • new releases of... Alceu Rodrigues de Freitas Junior via cpan-testers-discuss
    • Re: new re... James E Keenan
      • Re: ne... Alceu Rodrigues de Freitas Junior via cpan-testers-discuss
      • Re: ne... Slaven Rezic
        • Re... James E Keenan
        • Re... Alceu Rodrigues de Freitas Junior via cpan-testers-discuss
          • ... David Cantrell
            • ... Alceu Rodrigues de Freitas Junior via cpan-testers-discuss
              • ... Alceu Rodrigues de Freitas Junior via cpan-testers-discuss

Reply via email to