The stars aligned and all three of us managed to get together.
We mostly talked about PPCs, both in the general shape of the process, and specifically the latest proposal, PPC0034.
- Given how the PPC process has worked out in practice, we discussed how much sense it makes and whether it solves a problem we actually have. We agreed that the steering council - and Perl overall - would still benefit from having some sort of declared process by which people can suggest and discuss new features, as separate from implementing them.
- The process at the moment doesn’t align very well with existing practice; at the same time, existing practice is not particularly structured. Rather than trying to define a new process, we think it better to clarify the documented process to more obviously match what we actually do, and try to iterate that way.
- PPC0034 is concerned with refalias parameters in signatures. Both the
refalias and declared_refs features are still currently marked as experimental, though it is not immediately clear in their overview tickets why. We should clarify the status of these before we fully commit to PPC0034.
- That aside, the overall document of PPC0034 seems good and we’re happy to merge it as a basis for ongoing experiment and a trial implementation.
[P5P posting of this summary]
All three of us met.
- The refalias draft PPC on the mailing list is looking good. We encourage Gianni to turn it into a full PPC doc PR
- We still would like more automation around making real CPAN distributions out of
dist/ dirs. Paul will write an email requesting assistance on that subject specifically
- Briefly discussed the subject of the
meta module handling signatures with named parameters. Further discussion will continue on the email thread.
[P5P posting of this summary]
Just Paul and Leon this week. We discussed:
- The newly-added
signature_named_parameters experiment needs adding to perlexperiment.pod, and experimental.pm. Hopefully before 5.43.5 release on Thursday
- The
experimental dist would be easier to manage if it lived in dist/, but it would also be nice if we had more automated tooling to create real CPAN distribution tarballs out of those directories. Would help for making Module-Corelist updates every month as well.
- We await a full write-up of a PPC document following on from the “refalias in signatures” Pre-PPC discussion
[P5P posting of this summary]
Just Paul and Aristotle again.
We discussed the unlimited statement modifier chaining proposal. Both of us thought that unlimited chaining is a bad idea, though we agreed that being able to combine one loop and one conditional would on rare occasions make things slightly nicer. However we also agreed that in terms of conceptual complexity, only either allowing arbitrary chaining or disallowing chaining entirely is justifiable, as opposed to specifying and implementing rules to permit it but only in certain specific cases, especially considering the marginal benefit. So we settled on agreeing with Larry’s original design decision to disallow them.
[P5P posting of this summary]
Only Paul and Aristotle this time.
We had a small amount of helpful feedback on the named signature parameters PR. Paul wants to merge by the end of the week for the purposes of inviting more feedback, assuming no issues are raised in the meantime.
[P5P posting of this summary]
