|
Fire is step THREE! | github.com/perl6/toolchain-bikeshed | Channel logs: irclog.perlgeek.de/perl6-toolchain/today | useful prior art: metacpan.org/pod/CPAN::Meta::Spec Set by moderator on 11 December 2016. |
|||
|
07:13
domidumont joined
07:18
domidumont joined
|
|||
| nine | I think by now I've fulfilled all of jnthn++'s requirements for merging lexical_module_load. So it's just a matter of if we want to wait till after the release or not. | 08:00 | |
| lizmat | I would be in favour of before | 10:08 | |
| ugexe | I'd be in favour of before too - it would be a good thing to show off in the advent calender | 15:12 | |
| perlpilot | +1 from me for the same reason as ugexe :-) | 15:14 | |
| lizmat | I think consensus is now to do it just after the release | 15:27 | |
| and the advent calendar is already full anyway, no? | |||
|
15:31
tbrowder joined
15:41
FROGGS joined
|
|||
| ugexe | I could slip it into mine, since I'm briefly going over short-name/long-name for module lookup | 15:43 | |
| then again my post is on thursday (a few days before release) | 15:44 | ||
| er friday | 15:46 | ||
| lizmat | ugexe: maybe a switch could be possible ? | 15:58 | |
| ugexe | I can ask... should I query those on the 18th+? | 16:03 | |
| lizmat | well, that seems logical :-) | 16:05 | |
.oO( should stop identifying myself with Bones ) |
|||
| ugexe | nine: I'm thinking `method !matching-dist`, when given an identity that lacks an :auth<> and/or ver such as Foo::Bar, needs a way to first check if `$CALLER_COMPUNIT.distribution.name eq $spec.short-name` exists. If it does exist, then add `$CALLER_COMPUNIT.distribution.[ver&auth]` to the short-name before looking it up. | 16:28 | |
| So if a distribution like HTTP::UserAgent provides HTTP::Request it can use `use HTTP::Request;` (sans auth/ver) and never get a version provided by a different distribution (see: github.com/sergot/http-useragent/c...cb33b882). superseded-by could eventually be used to point at a HTTP::Request provided by a different distribution. | 16:29 | ||
| `return %!loaded{$spec.short-name} if %!loaded{$spec.short-name}:exists;` caching will likely need to be tweaked as well | 16:35 | ||
|
19:33
domidumont joined
19:34
domidumont joined
|
|||
| nine | ugexe: sounds like a policy. Like the Linux kernel developers I'd prefer to leave policy to the users and just provide mechanism. | 20:40 | |
| Or at least try to figure out what possible best practices we should have so that such a policy is not necessary normally. | |||
| For example, I'd still like our module authors to communicate openly and often about how to name modules and to avoid naming conflicts as much as possible. | 20:41 | ||
| CPAN has come very far with prohibiting re-use of names. It's just a hand full of cases where an auth mechanism would have made it easier for users and developers. I'd like this tool to be used sparingly and wisely as a kind of last resort. | 20:43 | ||
| My rationale being that a user installing HTTP::UserAgent will not expect to get an HTTP::Request, too. And I'd hate to see auth being used extensively just because authors are to lazy to split off commonly used or usable modules from their dists and working together on the basics. | 20:44 | ||
| lizmat | nine: +1 | 21:11 | |
| ugexe | true, but its also difficult to understand from a developer standpoint because the behavior would not be consistent across CURs. when I do -Ilib I get one distribution's version of HTTP::Request, and when I install the distribution it may end up using another | 22:57 | |