On Wed, Jul 17, 2024 at 2:49 AM Philippe Bruhat (BooK) <b...@cpan.org> wrote:
> On Mon, Jul 15, 2024 at 07:30:59PM -0700, Russ Allbery wrote: > > > > Some of it makes me wonder if I misunderstood the previous discussions > and > > the implications of the Lyon Amendment for the usability of CPAN tools > and > > for what prerequisites have to be met before bumping the minimum Perl > > version of way upstream packages to versions of Perl up to 5.16. I feel > > like I'm hearing different things from different people, all of whom are > > far more plugged in to the community than I am. > > One of the arguments during the discussion in Lyon was that continuing > to support old version like 5.10 implies that a lot of specialized > code needs to be kept to work around Perl bugs that have been fixed > decades ago. > > Another argument (I don't remember if it was made then) is that people > who refuse to update their code, insist on upgrading the environment > around it, and demand that their code keeps working unchanged, are > pushing the maintenance burden on everyone else but themselves. > > During that Toolchain Summit, there were also discussions of per-version > indices for CPAN, so that the CPAN clients could use an index tailored > for specific versions of Perl if the user wants them to. That way, those > who decide to live in the past don't have to be such a burden on those > living in the present and looking at the future. Sadly, little progress > has been made on this since. > > Such indices, combined with the "maximum minimum version" of 10 years > earlier sound to me like a good compromise. The toolchain gives you 10 > years; if you want/need more, use the old indices. > Yeah, such an index would remove a lot of the tension between these different interests, though it wouldn't eliminate all of it. We may need to be more proactive with recruiting someone (or some people) to actually set it up. Leon
