forum.polymake.org

There is a difference in perspective between a linux distros package maintainer and you as an upstream developer. For maintaining the package a hack that does its job for 2 distro releases is good enough. The alternative is not having polymake in any linux distro for (my estimate) two releases.

We tried to find such a hack but we failed. The problem is that the perl interpreter has hidden some vital functions which were necessary for parsing polymake perl code, starting with such elementary things as Matrix<Rational>. They can't be "revealed" from outside without modifying the perl executable itself (resp. its shared library). The only viable solution from the perspective of a package maintainer is to provide packages with perl 5.36 and a dozen of CPAN modules requiring binary compatibility (like Term::ReadLine::Gnu) in the corresponding older versions, and declare them as prerequisites for polymake package. python community used to provide packages with various versions of python 2 and 3 in parallel for quite a few years until venv became popular and they decided that all relevant modules finally managed to migrate to python 3. But obviously, it's a completely different level of importance which would justify such substantial maintenance efforts, that's not given for a small niche product like polymake.

Thanks for the reply. I have just found out the the Debian julia package is dead. So it is unlikely that we will have a polymake package in Debian in the long run and a short term solution is no longer important.

I feel sad about that, but while I would have been able to support a hack I don't think I can do anything about this. I am not even a Debian polymake maintainer. I became aware of the situation because I am maintaining the Debian gap-polymaking package which will also vanish. At that time I considered the vanishing of polymake from Debian as something I did not want to accept.

I want to thank you for releasing polymake 4.11. That is a much needed lifeline to help us distribution packagers keep things going for a little while longer. I wish you good speed on the Julia version.

Big thanks from me, too.

Well, it was primarily your insistence that forced me to have one more close look at that issue But the general warning remains: the current hack may cease to work with any new perl release.
BTW what happened to julia Debian package? Did they decide to move to snap, like firefox and chrome?

Asking that I got this answer on the Debian julia list.