Perl 6 - the future is here, just unevenly distributed

IRC log for #metacpan, 2014-08-21

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

All times shown according to UTC.

Time Nick Message
01:28 FROGGS__ joined #metacpan
01:40 klapperl_ joined #metacpan
03:19 metacpan joined #metacpan
03:19 metacpan [metacpan-client] kentfredric opened pull request #18: Dependency and Metadata fixes. (master...master) http://git.io/Ny9TMA
03:19 metacpan left #metacpan
03:20 dipsy [ Dependency and Metadata fixes. by kentfredric · Pull Request #18 · CPAN-API/metacpan-client · GitHub ]
03:26 talina_ joined #metacpan
04:10 mmp joined #metacpan
05:45 oiami joined #metacpan
06:19 madn3ss joined #metacpan
06:29 neilb joined #metacpan
07:49 neilb joined #metacpan
08:11 neilb joined #metacpan
08:51 dp_ joined #metacpan
09:21 talina_ joined #metacpan
09:36 oiami1 joined #metacpan
11:45 dp_ joined #metacpan
12:21 oiami1 left #metacpan
12:27 dp_ joined #metacpan
15:06 metacpan joined #metacpan
15:06 metacpan [cpan-api] oalders created oalders/module-indexing (+1 new commit): http://git.io/fZvtTQ
15:06 metacpan cpan-api/oalders/module-indexing f21555f Olaf Alders: Only add indexed modules to a file document....
15:06 metacpan left #metacpan
15:06 dipsy [ Only add indexed modules to a file document. · f21555f · CPAN-API/cpan-api · GitHub ]
15:07 oalders i didn't add a test case for that, but that appears to fix things
16:09 metacpan joined #metacpan
16:09 metacpan [cpan-api] oalders merged oalders/module-indexing into master: http://git.io/Ofaszw
16:09 metacpan left #metacpan
16:09 dipsy [ Comparing 19caae093f9f...f21555fa9cc6 · CPAN-API/cpan-api · GitHub ]
16:19 oiami joined #metacpan
16:29 theory joined #metacpan
16:45 neilb joined #metacpan
16:47 neilb_ joined #metacpan
16:51 neilb__ joined #metacpan
16:59 neilb_ joined #metacpan
17:44 rashi joined #metacpan
18:35 dpe_ joined #metacpan
19:31 neilb joined #metacpan
19:34 neilb ranguard: not that hard maybe, but still appreciated
19:35 ranguard :)
19:53 thaljef joined #metacpan
20:04 ether when I see this in my cpanm build.log:  Searching Moose::Cookbook::Legacy::Debugging_BaseClassReplacement () on metacpan ...
20:05 ether what query is it doing?  it's getting an old dist as a result (which does not match what's in PAUSE)
20:05 rwstauner dunno, that's a question for miyagawa
20:05 rwstauner or, read the cpanm source
20:06 ether *nod*, it's a hard slog so I was hoping someone would know :)
20:06 oalders yeah, he built those queries in consultation with us, but i can't remember exactly what they do
20:06 ether which would then let me try to figure out why metacpan is returning such a strange result
20:11 dp_ joined #metacpan
20:13 ether ok, I found the code here - https://github.com/miyagawa/cpanminus/blob/devel/lib/App/cpanminus/script.pm#L597
20:13 dipsy [ cpanminus/script.pm at devel · miyagawa/cpanminus · GitHub ]
20:13 ether I'm not sure if this is the specific query that's returning the bad result, though
20:14 ether or how I can construct this query to see what it would return (there's a query builder page somewhere, isn't there?)
20:14 oalders explorer.metacpan.org
20:24 ether wow, editing cpanm code is horrific
20:24 ether because of the minification
20:24 ether breathe on it funny and it no longer parses
20:50 rwstauner blue-face programming?
21:13 neilb joined #metacpan
21:15 neilb joined #metacpan
21:20 ether ok, I found the URL it's hitting...
21:20 ether http://paste.scsys.co.uk/417484
21:20 dipsy [ magnet_web paste from "ether" at 217.168.150.38... ]
21:21 ether now, is the query it's submitting correct, and if it is, why is that response being returned?
21:21 ether that's an old version of the dist, that is not in 02packages.details.txt anywhere.
21:30 ether this is the only package in the Moose dist that behaves this way, that I can see
21:31 ether I should probably open a ticket for this
21:32 ether as it probably implies something is weird in the metacpan indexer
21:33 trs ether: mitmproxy or tcpdump is my favorite way to debug cpanm
21:52 haarg ether: pm vs pod
21:53 ether haarg: it's pod now, but the change wasn't made around hte 2.0604 release.
21:53 haarg but the rest are on backpan
21:53 haarg 2.0604 is the latest on cpan with .pm files
21:55 ether but PAUSE says: Moose::Cookbook::Legacy::Debugging_BaseClassReplacement 2.1211  E/ET/ETHER/Moose-2.1211.tar.gz
21:56 ether so, metacpan is doing some extra thinking on top of what PAUSE says
21:56 rwstauner it is
21:56 ether I'll write up all this when my pants stop being on fire...
21:56 rwstauner api.metacpan.org/module/Moose::Cookbook::Legacy::Debugging_BaseClassReplacement shows the correct version
21:57 rwstauner however that cpanm query is specifically looking for the "modules" array
21:57 rwstauner and the metacpan indexer doesn't store "modules" on a pod file
21:57 rwstauner i was looking at this code just yesterday
21:57 rwstauner oalders: ^
21:57 rwstauner the file is linked to the module name b/c of the provides metadata
21:58 oalders ah, similar to what i patched this morning?
21:58 rwstauner but then the module info is not being saved on it (b/c it's a pod file, and pod files don't normally have packages)
21:58 rwstauner similar, but for a different reason
21:58 rwstauner there's a line of code that says `$file->clear_modules if $file->is_pod_file`
21:59 rwstauner so the question would be, is it correct to specify that a pod file provides a package?
22:00 oalders good question
22:00 rwstauner b/c technically it doesn't
22:00 oalders i just found this as well https://github.com/CPAN-API/cpan-api/issues/332
22:00 dipsy [ Coro.pm is provided by two different files · Issue #332 · CPAN-API/cpan-api · GitHub ]
22:00 oalders and this is sort of vaguely related
22:00 rwstauner you can't load the module, and there is no package declaration
22:00 rwstauner so, should the meta spec allow such a thing?
22:00 rwstauner or pause, in this case
22:01 rwstauner yeah, that stuff is trouble
22:05 oalders i would think only one of the modules should be indexed and authorized.  shortest match wins
22:07 rwstauner whichever one matches 02packages should win
22:07 oalders right
22:07 rwstauner there's some confusing language there, too
22:08 rwstauner module and file are the same
22:08 rwstauner so it looks like a module can have an array of modules
22:08 rwstauner really that array should be packages
22:08 oalders we could fix that in v1
22:08 rwstauner i believe the "definition" of a module is a file in INC that defines a package of the same name
22:09 rwstauner or something along those lines
22:10 oalders with the Coro problem, how do you know which one matches 02packages? 02packages just points to a tarball
22:11 oalders would be nice if 02packages provided a file name
22:11 oalders maybe 03packages would...
22:12 ether the root problem (or a contributor to it) could be that the Moose dist is explicitly adding meta 'provides' entries for all the pod
22:12 ether so PAUSE dutifully adds them to the index, whereas they probably shouldn't be indexed.
22:14 ether ..as they have no package declarations inside, nor any $VERSION assignment
22:14 ether I wonder if PAUSE should be validating the provides data it gets -- "bzzt, you gave me a 'provides' entry for $file and $package but that isn't in this dist!"
22:15 ether xdg: thoughts? ^^
22:15 ether ah, he's not here. foo.
22:15 ether name too short to use autocomplete
22:16 rwstauner :-)
22:16 rwstauner yeah, that would be my question... should pause really be indexing a pod file?
22:16 ether possibly not, as the point of 'provides' is to tell PAUSE what to index where it might be hard to scan the files to figure it out (e.g. a .pl file that creates a .pm at install time)
22:16 rwstauner or i guess, do people want it to?
22:17 rwstauner if that's the consensus (to just trust provides data) then we could certainly alter metacpan's indexer for consistency with that
22:17 haarg people want to be able to install documentation by name
22:17 ether .pod files can be valid perl, with a package and $VERSION declaration - in that case, they should be treated like a .pm (IMO)
22:17 rwstauner true
22:17 ether but in Moose's case, they're just docs, no perl at all.. so... it's a bit of a toss-up
22:17 haarg and people want pod docs to have permissions
22:18 ether rjbs: ^^ your opinion too please
22:18 ether at the very least, metacpan should be respecting meta provides just like PAUSE does
22:19 ether so (even if the provides data is wrong), if Moose's META.json says to index this .pod file as package X and version Y, metacpan should dutifully index it as such
22:19 haarg i don't see any value to excluding .pod files from indexing or treating provides as authoritative even if there is no resultant .pm file
22:19 ether then at least everything would be consistently wrong :), and the responsibility lies with the dist to get the metadata correct.
22:20 rwstauner yeah
22:20 haarg and putting a $VERSION / package declaration in a .pod file is just silly
22:20 rwstauner :-)
22:30 oalders theory: do you know when the next Pod::Simple release may be?
22:49 theory I don’t. Could be any time I guess.
23:28 vanstyn_ joined #metacpan
23:56 metacpan joined #metacpan
23:56 metacpan [cpan-api] oalders pushed 1 new commit to master: http://git.io/7zWj_Q
23:56 metacpan cpan-api/master 53956a4 Olaf Alders: Merge pull request #329 from CPAN-API/talina/documentation...
23:56 metacpan left #metacpan
23:56 dipsy [ Merge pull request #329 from CPAN-API/talina/documentation · 53956a4 · CPAN-API/cpan-api · GitHub ]

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