On Sat, Aug 11, 2018 at 4:39 PM Shlomi Fish <shlo...@shlomifish.org> wrote:
> Hi all! > > This post is a little flamebait, so please try to keep the discussion > civil. > > Anyway, after reading the discussion in this public github issue, and > following > some of the links (especially > > https://szabgab.com/what-does--if-it-aint-broke-dont-fix-it--really-mean.html > ), > do you think I was being unreasonable, or should I as a CPAN > author/maintainer/adopter accommodate for people running old perl5s, in my > case > 5.10.x and below: > > https://github.com/shlomif/perl-XML-SemanticDiff/issues/3 > > This reminds me of what chromatic wrote here - > https://www.nntp.perl.org/group/perl.perl5.porters/2008/09/msg140206.html > : > > « > This is why we can't have nice things. > » > > Any comments or opinions? I think I'll relax by watching a nice and fun > video. > > Here is my take on the issue. 1. There is a general expectation that old but supported (by toolchain) Perls will continue to work, but perhaps not be able to use new features. This is just the legacy of Perl. 2. As a CPAN author you have no obligation to support anything you don't want to. Users have the ability to fork your module or use a different one if they need to. 3. It's generally polite to your users to declare and support a minimum Perl version that is consistent with what Perl features you are using, and that is simple for you to test on to ensure continued support (e.g. via Travis CI or cpantesters, if not your own system). With all of that combined, my personal policy for CPAN modules that I expect others to use is to support back to 5.8 unless it is in an ecosystem that already requires a newer version (like a Mojolicious module requires 5.10, a Dist::Zilla module requires 5.14). But that is just because that is the minimum version which is supported by toolchain and convenient for me to test through Travis CI. Your support policy can be whatever you want it to be, but be aware how people will decide to go about using or not using your module as a result. -Dan
