|
Fire is step THREE! | github.com/perl6/toolchain-bikeshed | Channel logs: irclog.perlgeek.de/perl6-toolchain/today Set by moderator on 9 January 2016. |
|||
|
02:49
ilbot3 joined
|
|||
| moderator | Fire is step THREE! | github.com/perl6/toolchain-bikeshed | Channel logs: irclog.perlgeek.de/perl6-toolchain/today | ||
|
11:58
woolfy joined,
woolfy left
12:01
woolfy joined,
woolfy left
12:20
sivoais joined
14:18
leont joined
14:28
sjn joined
|
|||
| mst | lizmat: um, I never said anybody was trying to destroy anything | 15:09 | |
| lizmat: I specifically called out it as a communication problem | |||
| lizmat: and I'm trying to get the fixes in, just in a co-ordinated fashion | |||
| lizmat: I'm tired of you jumping at random to anti-perl5 rants when I'm trying to stand in the middle and keep everybody happy here; if you want to yell at somebody, find somebody who's actually against you, I'm sure there's somebody somehwere | 15:10 | ||
| meanwhile, the proximate thing that went wrong was the trial PSIXDISTS uploads switched from dev releases (which most things skip) to "real" releases (which most things don't) via a two line conversation on #metacpan that it appears both of the participants in misunderstood the purpose and results of | 15:11 | ||
| I got those two people talking originally. I've invited them both into here. And I think their effort is probably going to be the right thing in the long run, and I -do- believe that using the cpan file distribution mechanism will be at least *a* useful mechanism of distributing perl6 modules, even if there are likely to be others as well | 15:13 | ||
| what I'm talking about is making sure we do it right, and trying to minimise *unexpected* breakage for people downstream of what we're doing, and try and make sure the people who'll see expected breakage know about it in advance | 15:14 | ||
| mst still suspects that having the primary non-dev upload feed filter out Perl6/ would quite possibly be sufficient, though I'd like to double check that with people | 15:29 | ||
| lizmat: if there's sections about using cpan for perl6 in the Lancaster or Berlin Consensus documents, could you point them out to me? Those might be useful for rolling up and thwapping people with who don't remember :) | 15:49 | ||
| sjn | \\o | 16:53 | |
| sjn reads the irclog | 16:54 | ||
| happy to see a channel for working on toolchain stuff :) | |||
| mst | lizmat: thinking about this, I think maybe the best answer is to have the current pause feed that everybody assumes is a perl5 feed exclude the Perl6/ directory, then mint two new feeds - one for Perl6/ stuff and one combined. then we can all have what we want without confusing existing downstream consumers. does that seem like a sane idea to you? | 17:01 | |
| mst is happy to propose it to modules@/andk and shepherd it through if so | |||
| sjn has some notes that he and tadzik jotted down last year. anyone interested in this braindump? :) | 17:03 | ||
| mst | absolutely. | 17:04 | |
| sjn | lots of bikesheddingthere :-P | ||
| pad.hackeriet.no/p/p6-deploy-guidelines (with links to related braindumps) | 17:05 | ||
| github.com/sjn/perl6-example-dist (initial attempt at making a "complete" example distribution, with all kinds of issues to handle) | 17:06 | ||
| mst | this looks like it's a worthwhile document to have to inform further panda/zef/etc. development | ||
| sjn hoped to have worked a bit more on this with tadzik, but things didn't turn out well for that | 17:07 | ||
| sjn toddles off to the flight to CPH | |||
| back in a few hours | |||
| mst | sjn: I'm going to add you to the bikeshed repo, throw this in there somewhere please | ||
| ugexe | should we change (or add a new similar) license field in the meta spec that is not a url but a name that can easily be grepped? | 17:15 | |
| mst | handling the Software::License names would be nice | 17:16 | |
| moderator | 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 | 17:16 | |
| mst | here we go, this is the perl5 version - metacpan.org/pod/CPAN::Meta::Spec#license | 17:17 | |
| I wonder if allowing, say, license:foo for a name, and then a URL for anything else, might be neat | |||
| autarch | If the system doesn't involve generating a Makefile that contains portable shell to execute p6, I'm not sure it wlil be a _real_ system | 17:18 | |
| mst | ugexe: just some thoughts, no particular recommendation | ||
| autarch: the CPAN::Meta spec might be useful prior art, but the makemaker source code not so much | |||
| ugexe | im trying to incorporate it into the config like bsd ports, but any way to white/black list is good in my book | ||
| autarch | obviously I was joking - yes, I think there's a lot to take from CPAN::Meta | ||
| mst | although admittedly MY:: in EUMM is the first ever instance of runtime role application in the wild so :) | ||
| leont | Yeah, MM is not something I'd like to copy | 18:42 | |
| mst | I do think 'being able to do the moral equivalent of "make -n"' is worthwhile | 18:44 | |
| I don't think any of the implementation is ;) | |||
| nine | One real huge adavantage Perl 6 has is that the final copying of files is under control of a core class. That's why (low level) uninstall will be so easy to implement | 18:45 | |
| mst | I'm mostly interested in 'make -n' for being able to debug compilation issue. I guess the idea is NativeCall means you shore those off to a separate package | 18:47 | |
| leont | Same is true in p5, the code class (ExtUtils::Install) just happens to suck | 18:48 | |
| s/code/core/ | |||
| lizmat | mst: github.com/Perl-Toolchain-Gang/too...nsensus.md nor github.com/Perl-Toolchain-Gang/too...nsensus.md contain anything about Perl 6 | 18:56 | |
| mst | lizmat: right. so apparently we don't so much have a communication breakdown as a russian doll made of layers of communication breakdowns | 18:57 | |
| lizmat | why? because everybody was busy with the Perl 5 toolchain | ||
| mst | and because presumably none of the perl6 attendees added their discussions to the document as it was being put together, probably due to yet another communication breakdown somewhere | ||
| lizmat | the *only* reason I attended the Lancaster, Lyon, Berlin QA hackathons, was to make sure we would have some kind of communication about Perl 6 | ||
| mst | yeah, well, next one, let's make sure the perl6 stuff actually makes it into the minutes | 18:58 | |
| lizmat | well, my feeling at the time, was that it was considered something that was not important | ||
| sounds like a plan to me | |||
| meanwhile, I should stress that FROGGS++ has spent quite some time in the PAUSE code to make uploads to PAUSE possible without breaking anything | 18:59 | ||
| mst | yeah, well, after this long, I don't think people really believed anymore, which while unfortunate is, given the timelines involved and previous messaging they'd heard from the perl6 project, quite understandable | ||
| lizmat | and since the pluggability of the PAUSE code is below zero, I think that was quite a feat | ||
| mst | I am annoyed by the situation but I don't think it was really anybody's fault exactly | ||
| and I think with the RSS feeds tweaked in the way I describe, we can experiment all we like without causing any other unexpected complications for downstream consumers | 19:00 | ||
| lizmat | well, you touched a nerve when you said "and since nobody talked to the perl5 toolchain team first" | ||
| mst | well, I went through the #toolchain logs which is how you normally do that and couldn't find anything significant since 2009 when moritz was saying "we should totally collaborate, but we're not sure what our side needs yet" | 19:01 | |
| lizmat | because trying to talk to the perl5 toolchain team is the reason I have attended the past 3 QA hackathons to begin with | ||
| mst | right, and then none of your efforts made it into any of the write ups so there was no way for me to tell that had even happened | 19:02 | |
| lizmat | and I can tell you, despite the good people there, I can think of nicer places to be hacking | ||
| nine | Oh come on, the food was great :) | ||
| tina++ tina++ | |||
| mst | well, yeah, you haad the year of schwern ranting at everybody, and the year of riba ranting at everybody ... | 19:03 | |
| anyway. one of the reasons I'm here is to try and make this go more sensibly | 19:05 | ||
| lizmat | nine: funny you should say that | ||
| mst | I know how to wrangle that bunch of crazy masochists just fine | ||
| lizmat | if you're there... :-) | ||
| mst | and that's why I'm going to do this on #toolchain, where I am always there :D | ||
| leont did way less p6 toolchain in Berlin than he had planned to | 19:06 | ||
| mst | I want 'uploads to PAUSE possible without breaking anything' to be true, including where possible downstream values of 'anything', because that's how toolchainery works | ||
| (actually, one of the reasons I'm not a huge fan of hackathon type events is *because* of stuff not getting written down, thereby resulting in disconnects like this that bite us on the arse later) | 19:08 | ||
| lizmat: so, on that basis, what do you reckon to the three feeds idea? | 19:09 | ||
| lizmat | I was not aware that there was such a thing as a "feed" | 19:10 | |
| nine | lizmat: and of course Wendy++ Wendy++ :) | ||
| lizmat | I was given the impression it was basically all rsyncs | ||
| mst | you're confusing PAUSE and cpan | ||
| lizmat | ah, could be :-) | ||
| mst | cpan, the giant distributed filesystem, is rsyncs or the moral equiavlent thereof | ||
| PAUSE, the server that accepts uploads and adds them to the master copy of the cpan, provides a feed of recent uploads | 19:11 | ||
| so for example freenode #perl sees every upload to CPAN announced | |||
| lizmat | mst: anyways, 3 feeds sounds like a plan to me | ||
| mst | #toolchain sees things like ExtUtils:: and Module:: | ||
| lizmat | and have the Perl 6 only feed report to #perl6 | ||
| flussence | ugexe: (2h ago) before we go off inventing new license shorthand (which I agree would be nice), has anyone heard of SPDX? It's supposed to standardise all that (in a way dumb source scanners can agree with) | ||
| mst | and downstream packagers use it to see what's changed so they can keep their packages up to date | 19:12 | |
| all of these users became a trifle confused when the PSIXDISTS uploads began | |||
| lizmat | wrt to writing things down: might I point out that without Wendy making the miinutes of both the Lancaaster and Berlin Consensus, there wouldn't have been much to communicate | 19:14 | |
| mst | lizmat: yeah, having it easy for us to do that seemed like rather a win anyway | ||
| (wrt the #perl6 upload feed idea) | |||
| might I point out that my only real comments here are that (a) I wasn't there so I don't know (b) that means if it wasn't written down - for whatever reason - I didn't have a realistic way of knowing | 19:15 | ||
| I don't think trying to figure out whose 'fault' it was/wasn't is productive, I think we just need to make sure that if it's something people could be blindsided by, it gets written down somewhere in future | 19:16 | ||
| ranguard | yea, especially for who ever uplaoded all those PSIXDISTS, and I view that as a way of testing where things need reviewing ;) | 19:18 | |
| lizmat | mst: I will make sure there will be something written down the next QA hackathon :-) | 19:20 | |
| mst dumps the idea in #toolchain for discussion before I file it as an issue against PAUSE | 19:22 | ||
| 19:24 <@klapperl> wfm | 19:25 | ||
| created github.com/andk/pause/issues/198 to track it | 19:40 | ||
|
19:42
hankache joined
|
|||
| mst | ranguard: ^^ you might have some useful input on this | 19:46 | |
| ranguard | mst: other than github.com/CPAN-API/cpan-api/blob/...#L127-L130 that's all the involvement I've had so far | 19:58 | |
| mst | right | ||