Perl 6 - the future is here, just unevenly distributed

IRC log for #metacpan, 2014-05-17

| Channels | #metacpan index | Today | | Search | Google Search | Plain-Text | summary

All times shown according to UTC.

Time Nick Message
01:12 klapperl joined #metacpan
04:07 ether_ joined #metacpan
04:12 talina_ joined #metacpan
04:14 oiami joined #metacpan
04:24 oiami joined #metacpan
07:07 neilb joined #metacpan
07:14 ether_ joined #metacpan
08:08 neilb joined #metacpan
08:31 neilb joined #metacpan
09:12 neilb joined #metacpan
10:09 oiami joined #metacpan
10:32 lifeofguenter joined #metacpan
11:26 lifeofgu_ joined #metacpan
12:02 bowtie joined #metacpan
12:08 neilb joined #metacpan
12:38 neilb joined #metacpan
14:06 lifeofguenter joined #metacpan
14:53 heytitle joined #metacpan
14:55 heytitle1 joined #metacpan
15:01 heytitle_ joined #metacpan
15:03 heytitle left #metacpan
15:03 heytitle_ left #metacpan
15:03 heytitle joined #metacpan
15:06 heytitle joined #metacpan
15:20 heytitle joined #metacpan
16:37 neilb joined #metacpan
16:46 * rwstauner thinks marking dists as UNAUTHORIZED will lead to many more people discovering that they have 'package DB' statements and no correlating "no_index" directive
16:50 mst this is why marking things unauthorized that have some permissions but not all is an error
16:50 mst it was a complete fucking disaster when s.c.o did that
16:51 rwstauner i thought we had that, but turned out we didn't, so i turned it on partially to see how much would be affected
16:52 rwstauner mst: have any input on https://github.com/CPAN-API/metacpan-web/issues/1164 ?
16:52 neilb joined #metacpan
16:52 dipsy [ bigint search · Issue #1164 · CPAN-API/metacpan-web · GitHub ]
16:54 mst rwstauner: does that comment clarify?
16:58 rwstauner that makes sense for the DB case, but i think that wouldn't change anything in the case of bignum
16:58 mst rwstauner: it'd mean rafl's was marked as latest
16:59 mst rwstauner: since that's what you'd get when you 'cpan bignum'
16:59 rwstauner right, that's the desire
16:59 mst exactly
16:59 heytitle I'just cloned metacpan-developer and tried to update dependency by using /home/metacpan/bin/install_modules.
16:59 mst but it'd still mean that 'no perms to DBIx::Class::Storage::DBI::Replicated::OopsNewNamespace::LastReleaserForgotToGiveMePerms' wouldn't be a disaster
17:00 rwstauner how so?
17:00 rwstauner for the dist with 1 unauthorized module, we're basically ignoring it
17:00 mst yes, and you're going to change that.
17:00 mst that's what the heuristic is for
17:00 rwstauner i've got something backwards
17:00 mst it means rafl's will still be marked as latest, without fucking everybody else
17:01 heytitle and It doesn't seem to work correctly https://gist.github.com/goonnow/72ab1e194f7c2a84f130
17:01 dipsy [ gist:72ab1e194f7c2a84f130 ]
17:01 mst rwstauner: basically, I am proposing that instead of 'authorized for everything' being the restriction for latest
17:01 mst rwstauner: you instead use 'authorized for join "::", split "-", $distname'
17:02 mst rwstauner: so that would correctly catch the bignum problem, but without inflicting false-positives on otherwise normally installable dists
17:02 rwstauner oh, ok
17:03 rwstauner now i see the difference you're suggesting
17:03 mst it's not the 'package DB' problem that first occurred to me, it's the many-releasers dists that constitute at least half of metacpan's dependency chain
17:03 rwstauner bignum *would* be correct, but Crypt-SSLeay wouldn't have had a problem
17:04 rwstauner yeah i know, i've been a part of that
17:04 rwstauner here, have half the perms!
17:04 mst so a new namespace can easily not get co-mainted right to begin with
17:04 rwstauner yep
17:05 rwstauner heytitle: did you do 'vagrant provision'
17:05 heytitle yep
17:05 mst note: search.cpan.org first introduced its UNAUTHORIZED stuff during a DBIx::Class dev release cycle ... and it created its own internal index of 'dev perms' based on first uploader ... except there was no way to assign those, because s.c.o internal black box
17:05 rwstauner then you shouldn't need to do install_modules by hand i don't think
17:05 rwstauner right
17:05 rwstauner interesting
17:05 mst so every time I had a new namespace in a devrel with a different uploader I got deeper into pain
17:06 mst I eventually got gbarr to rewrite most of the logic based around reality as it was rather than reality as cpan used to be before the cabals became a thing
17:06 mst but it means I got to see lots of 'fascinating' failure modes
17:07 rwstauner fun
17:09 heytitle rwstauner: here the reason https://gist.github.com/goonnow/f92746a9e6e0296d9508
17:09 dipsy [ gist:f92746a9e6e0296d9508 ]
17:21 mst rwstauner: thank you for the explanation for humans of what I meant :)
17:21 rwstauner :-)
17:22 rwstauner i just had to write it down again for myself
17:22 mst I think between the two comments it should now be pretty clear
17:22 mst and pretty obviously the right thing to do
17:22 mst (or at least, 'least wrong' ;)
17:25 rwstauner :-)
17:29 rwstauner heytitle: try just doing `/home/metacpan/bin/install_modules Data::Pageset` directly
17:30 rwstauner there's something going on with cpanm that i have to figure out
17:30 heytitle thanks!
17:31 rwstauner heytitle, can you gist the output of 'ls -al /home/metacpan/metacpan.org' ?
17:32 heytitle rwstauner: https://gist.github.com/goonnow/22d834b389a61b58e8a5
17:32 dipsy [ gist:22d834b389a61b58e8a5 ]
17:33 rwstauner heytitle: thanks, that answers my question, though it's not what i was expecting to see
17:33 heytitle aha
17:33 rwstauner the issue is that cpanm doesn't seem to be reading the cpanfile correctly
17:33 rwstauner oh, nm
17:33 rwstauner that *is* what i expected to see
17:34 rwstauner on my vm, i have a Makefile and MYMETA files, so I get a big list of modules that it's going to install, but Data::Pageset isn't in it
17:34 rwstauner b/c those files are out of sync (I should delete them)
17:34 heytitle ic
17:34 rwstauner but in your output there was no list at all, so that makes sense
17:34 rwstauner but if i do the cpanm command that the wrapper should be doing, i get the right list
17:34 rwstauner so i'm very confused now
17:36 rwstauner aha
17:36 rwstauner there's some sort of disjoint with the PATH happening, which seems weird
17:37 heytitle :|
17:37 rwstauner wait, not PATH, but some other env vars
17:38 rwstauner the problem goes away if i source a bunch of env vars
17:38 rwstauner but there are more vars in there than i realized
17:38 rwstauner strange
17:40 rwstauner nm
17:40 rwstauner that's not right either
17:50 rwstauner heytitle: do 'git pull' on your metacpan-puppet repo, then do 'vagrant provision' in the developer repo, then  try that install_modules command... you should see a big list of modules now (where your previous gist showed N/A and no list of mods)
17:56 heytitle aha
18:04 rwstauner now that we have cpanfile everywhere i could probably rewrite most of that wrapper script
18:04 rwstauner anyway, thanks for pointing that out
18:06 heytitle you're welcome
18:07 metacpan joined #metacpan
18:07 metacpan [metacpan-developer] rwstauner pushed 2 new commits to master: http://git.io/bR7gRg
18:07 metacpan metacpan-developer/master 2a4cd0b Randy Stauner: Add Data::Pageset to the list of mods to install
18:07 metacpan metacpan-developer/master bb978a5 Randy Stauner: Add a TODO comment about checking each repo for mods
18:07 metacpan left #metacpan
18:07 dipsy [ Comparing af3b5eb040dd...bb978a5bd8d7 · CPAN-API/metacpan-developer · GitHub ]
18:12 metacpan joined #metacpan
18:12 metacpan [metacpan-examples] Talina06 opened pull request #7: Changes to Author examples. (master...APIchanges) http://git.io/JnE2fw
18:12 metacpan left #metacpan
18:13 dipsy [ Changes to Author examples. by Talina06 · Pull Request #7 · CPAN-API/metacpan-examples · GitHub ]
19:10 SineSwiper joined #metacpan
21:52 neilb joined #metacpan
22:51 jnbek joined #metacpan
22:57 jnbek joined #metacpan

| Channels | #metacpan index | Today | | Search | Google Search | Plain-Text | summary