We finally managed to arrange our first regular meeting between the three of us.

• Largely we discussed strategy for the named parameters branch. We agreed to merge soon (so people start playing with it), just staying ready to back it out well before release, in case it proves not to be ready.
• In that context, we considered the situation with experiment warnings. We agreed that named parameters should be an additional experiment of its own, though it is an adjunct of another experiment – and we do not yet have established ways of dealing with such a situation. We want to think about how our experiment mechanism should be extended to cover such cases.

[P5P posting of this summary]

Ongoing scheduling issues have meant we haven’t met all three together for a while, but today Paul and Leon found a time to discuss a few issues.

Dev point releases

5.43.3 just went out. .4 to .7 are accounted for, so we’ve a few months yet on that. Nothing for PSC to do for now.

OpenSSL progress

Leon has been building a replacement for Net::SSLeay - an XS wrapper of libssl. Eventual plan is that IO::Socket::SSL should be able to use that instead. Eventual intention is that it can be bundled with the actual core perl dist and get us ability to use https URLs from the in-core CPAN client directly.

We should have a technical meeting at some point to round up some of the interested parties. PSC’s involvement can just be keeping an eye on it, and seeing if it is on track to be included in (5.)44.

TODO: Plan a time and audience for said meeting.

Better categorization of core vs dual-life modules

There’s probably more modules in ext/ than there ought to be (e.g. File-Find). It’d be nice to tidy these up and move them out into cpan/ or dist/.

TODO: Leon will email p5p@ to form a list of suggestions

In-core module modernizing

Started by JRaspass, we have been modernising some core modules and unit tests, by ensuring they begin use v5.40; This has mostly gone fine, and is a thing we wish to encourage and invite more people to do with more files.

Paul encountered a problem with attributes.pm (well, .xs) wherein it picks up a warning from the wrong caller layer. It’s a bug that can be fixed but it might provoke illegalproto warnings in code that previously didn’t. But this is probably fine in the general category of “eh, new perls might warn about new stuff” so probably fine.

We’ll carry on doing more module modernisations, but we should wait on the results of better categorization (see above) first.

“Signatures Named Parameters”

Paul wants to merge the signatures-named-parameters branch sometime soon, because it’s been sat for 2 months with no comments. Will set a “will merge on this date” deadline - comments would be most appreciated by then.

One remaining thought is whether this new ability should provoke expermental warnings? If so, in what category?

In general we don’t actually have a policy, or any guiding historic precedent, on how to continue an experiment once it (or parts of it) is declared stable.

TODO: Ask p5p@ what we think of this expermenting plan.

“English Names”

There’s been two previous suggested implementors for this, but no code has ended up materialising. I (Paul) will probably just have a go at writing it myself so then it will be done. It would have been a useful mentoring/teaching opportunity, but since we’ve already lost two people by attempting it, I don’t want to risk a third ;)

[P5P posting of this summary]

All three of us attended, but with Aristotle and Paul short on time. So this was a short meeting. We discussed some administrivia and reviewed the left-over todo list from the previous PSC.

[P5P posting of this summary]

The transition meeting to the new PSC proved a bit tricky to schedule to get everyone from both the old and new PSC in attendance, but eventually we succeeded: Aristotle, Graham, Leon, Paul, and Philippe all participated.

• We discussed our structure for PSC meetings and our learnings about it from last cycle. We briefly introduced Leon to it and went over the onboarding checklist.
• We discussed roadmap items from the last cycle that did not get done, such as getting TLS in core. No decisions were taken as they will be for the new PSC to make.

[P5P posting of this summary]