Front page | perl.perl5.porters |
Postings from April 2016
* Ricardo Signes <perl.p5p@rjbs.manxome.org> [2016-04-08 20:23]: > Todd Rinaldo suggested that maybe we should stop treating IO as > dual-life, as it might break things if released now, and that it might > make more sense to only get a new IO(::File) when you upgrade perl. A release *might* break things, but un-dual-life’ing is *certain* to: http://grep.cpan.me/?q=file%3AMETA.json+%22IO%28%3A%3A%28Dir%7CFile%7CHandle%7CPipe%7CPoll%7CSeekable%7CSelect%7CSocket%28%3A%3A%28INET%7CUNIX%29%29%3F%29%29%3F%22%5B+%5Ct%5D*%3A%5B+%5Ct%5D*%22%28%5B%5E0%5D%7C0%2B%5B%5E%22%5D%29 The fact that dists may take dependencies on dual-life modules basically implies that dual-life modules must never become indexed under the perl dist again. I suppose exceptions to this rule are conceivable on a case-by-case basis, but that seems a very long shot in practice. The module at hand would have to have extremely few revdeps, and due to lack of visibility into the DarkPAN even this is probably only a reasonable criterion if the module was dual-life’d very recently. (Which of course would raise the question of why nobody saw the problem coming (whatever it was that would cause a desire to re-un-dual-life the module).) Basically: once dual-life, always dual-life. > If we want to get a new IO dist released, we should just do it: make > a dev release (there have been some over the past years), test it, and > release it after a test period. +1 Regards, -- Aristotle Pagaltzis // <http://plasmasturm.org/>Thread Previous | Thread Next