Camelia, the Perl 6 bug

IRC log for #metacpan, 2013-05-07

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

All times shown according to UTC.

Time Nick Message
01:10 thaljef joined #metacpan
01:16 SineSwiper joined #metacpan
01:49 cooper joined #metacpan
03:14 thaljef joined #metacpan
03:16 ether_ joined #metacpan
03:44 preflex_ joined #metacpan
04:06 gshank_ joined #metacpan
04:46 ether joined #metacpan
05:09 lucs joined #metacpan
05:11 lucs Trying to run tests for cpan-api, and the server config file has "cpan t/var/tmp/fakecpan", but 'fakecpan' is actually in "t/var/fakecpan". Is this as it should be?
05:12 * lucs goes to bed, will check back tomorrow.
05:20 rwstauner lucs: yes.
05:58 metacpan joined #metacpan
05:58 metacpan [metacpan-web01] oalders pushed 1 new commit to master: http://git.io/D9eLrQ
05:58 metacpan metacpan-web/master a007a79 Carwyn Ellis: Issue#671...
05:58 metacpan left #metacpan
05:58 dipsy [ Issue#671 · a007a79 · CPAN-API/metacpan-web · GitHub ]
06:01 metacpan joined #metacpan
06:01 metacpan [metacpan-web01] oalders pushed 2 new commits to master: http://git.io/guI9ag
06:01 metacpan metacpan-web/master bd3718f Carwyn Ellis: Issue#697...
06:01 metacpan metacpan-web/master 56a4f1a Carwyn Ellis: Issue#697...
06:01 metacpan left #metacpan
06:01 dipsy [ Comparing a007a7946cca...56a4f1a01738 · CPAN-API/metacpan-web · GitHub ]
06:05 metacpan joined #metacpan
06:05 metacpan [metacpan-web01] oalders pushed 3 new commits to master: http://git.io/ZzbaQQ
06:05 metacpan metacpan-web/master dc7582f Carwyn Ellis: Issue#711...
06:05 metacpan metacpan-web/master 72e8dd8 Carwyn Ellis: Issue#711...
06:05 metacpan metacpan-web/master dc87c12 Carwyn Ellis: Issue#711...
06:05 metacpan left #metacpan
06:05 dipsy [ Comparing 56a4f1a01738...dc87c1295938 · CPAN-API/metacpan-web · GitHub ]
06:35 Mike-PerlRecruiter_ joined #metacpan
06:43 dpetrov_ joined #metacpan
08:05 przemek joined #metacpan
08:36 daxim joined #metacpan
08:51 przemek joined #metacpan
08:57 przemek left #metacpan
09:30 nperez left #metacpan
10:58 przemek joined #metacpan
11:02 przemek left #metacpan
11:23 haarg joined #metacpan
11:26 bowtie joined #metacpan
11:27 daxim joined #metacpan
12:35 clkao left #metacpan
12:51 Amrita joined #metacpan
12:54 lucs rwstauner: Hmm... Okay. So the tests are failing for some other reason I guess. I'll try to figure it out.
13:20 GumbyNET5 joined #metacpan
13:36 metacpan joined #metacpan
13:36 metacpan [metacpan-web01] monken pushed 1 new commit to master: http://git.io/7iRjtA
13:36 metacpan metacpan-web/master f6eefd2 Moritz Onken: use fq link if file is not authorized or indexed
13:36 metacpan left #metacpan
13:36 dipsy [ use fq link if file is not authorized or indexed · f6eefd2 · CPAN-API/metacpan-web · GitHub ]
13:39 metacpan joined #metacpan
13:39 metacpan [metacpan-web01] monken pushed 1 new commit to master: http://git.io/Vfo43g
13:39 metacpan metacpan-web/master 2145c37 Moritz Onken: fetch indexed and authorized fields
13:39 metacpan left #metacpan
13:39 dipsy [ fetch indexed and authorized fields · 2145c37 · CPAN-API/metacpan-web · GitHub ]
14:17 ether very nice changes today Olaf! thank you thank you!
14:18 przemek joined #metacpan
14:19 wreis joined #metacpan
14:19 przemek left #metacpan
14:25 rwstauner lucs: feel free to describe the problems you're seeing and we'll try to help figure them out
14:26 rwstauner lucs: t/var/fakecpan is the source data that generates the fakecpan into t/var/tmp/fakecpan
14:26 rwstauner when t/fakecpan.t is run
14:35 lucs Ah, I see.
14:35 lucs Let me nopaste the error I'm getting...
14:39 lucs http://scsys.co.uk:8002/246653
14:39 dipsy [ freenode_web paste from Someone at 70.81.138.180... ]
14:40 rwstauner the error is from ES which is not related to the cpan dir on disk
14:41 lucs Hmm...
14:41 rwstauner ES has its own index in its own dir
14:41 rwstauner what command did you run to start ES?
14:41 lucs I'll nopaste it...
14:42 lucs http://scsys.co.uk:8002/246654
14:42 dipsy [ freenode_web paste from Someone at 70.81.138.180... ]
14:43 lucs And the .yml file only has "path" settings that point to a directory for the ES instance.
14:44 rwstauner hmm. ok
14:44 rwstauner any interesting error output from ES?
14:46 wreis joined #metacpan
14:47 am0c joined #metacpan
14:48 lucs Only that perl test output as far as I can tell. And the ES log adds these lines: http://scsys.co.uk:8002/246655
14:48 dipsy [ freenode_web paste from Someone at 70.81.138.180... ]
14:49 lucs And it appears that ES's ".../testing" subdirectory was touched, many files appears to have been touched.
14:50 lucs Actually, it's the .../testing/nodes/0/indices that has touched files.
14:51 rwstauner oh, well there are 'create_mapping' lines in the 'cpan_v1' index which is what you'd want to see
14:52 lucs You mean, that part is expected, and good?
14:52 rwstauner indeed
14:52 rwstauner that's what my output looks like
14:52 rwstauner what version of ES do you have?
14:52 lucs 0.90 something I believe.
14:53 lucs The most recent that was available.
14:53 rwstauner right
14:53 rwstauner i think metacpan is using 0.2x.x (0.9 is a beta)
14:54 rwstauner not sure if it would make any different
14:54 rwstauner s/t$/ce/
14:54 lucs Hmm...
14:54 rwstauner i haven't tried 0.90 yet
14:54 lucs I can try downgrading.
14:54 rwstauner on my ES the next lines i see are for deleting and creating the "user" index
15:03 lucs Aha. With 0.20.6, that test succeeds.
15:03 rwstauner good to know
15:03 lucs N'est-ce pas :)
15:03 rwstauner i read the release notes on 0.90 once but didn't pay attention to what might break
15:03 rwstauner i know they added compression or something
15:03 rwstauner not sure what needs to happen to work with that
15:04 lucs Thanks for your help -- I'm okay with 0.20.6, it's just to play with metacpan.
15:04 rwstauner np
15:04 rwstauner glad to have you fiddling with it :-)
15:04 rwstauner have fun
15:04 lucs Thanks :)
15:06 lucs By the way, you might want to mention the 0.90 current incompatibility on https://github.com/CPAN-API​/cpan-api/wiki/Installation
15:06 dipsy [ Installation · CPAN-API/cpan-api Wiki · GitHub ]
15:07 rwstauner lucs++ i was planning on adding an issue or something so that we at least document the erroneous behavior :-)
15:07 rwstauner am on a call now
15:09 daxim rwstauner, I'm kinda annoyed from your reply on issues/835 - don't play coy, if you know how to predict the permalink for a distro, then tell me straightaway
15:15 rwstauner daxim: didn't mean to annoy, i just don't know the details of your specfile and/or deploy system. the URL is predictable: /${\substr($author, 0, 1)}/${\substr($author, 0, 2)}/$author/$release-$version
15:15 rwstauner i have no idea how to get that into whatever template thing you have
15:15 rwstauner but assuming you know who's releasing it, you could have that PAUSE id data
15:16 daxim gosh.  okay, what's the author for DBIx-Class-0.08251?
15:16 rwstauner whoever is releasing it at the time?
15:16 daxim you cannot say "whoever is releasing it at the time" in a specfile
15:16 rwstauner i'm sorry, i don't know what specfile you're referring to
15:17 rwstauner I release with Dist::Zilla which generates files at release
15:17 rwstauner so I can do that
15:17 rwstauner i have no idea what system you're referring to
15:17 daxim you're frustrating me to no end
15:17 rwstauner i'm sorry, i don't mean to
15:18 rwstauner but you haven't given me any details of what this 'specfile' is either
15:18 daxim I explained already well enough that the permalink (minus version number) changes all the time
15:19 rwstauner yes, but i don't understand why the version is allowed to change and other parts of the url are not
15:19 rwstauner you could just tell me "it doesn't work that way"
15:19 rwstauner but i still have no idea what this 'specfile' is or where it comes from
15:32 Amrita joined #metacpan
15:33 mst daxim: you need to track module versions not dist versions, I think
15:34 daxim nah, it's for the archive permalink ("Source:")
15:40 haarg daxim: the release filename on its own is not unique.  DBIx-Class-0.08250 on its own isn't enough information to generate a permalink.
15:42 daxim I've read the linked bug report and to me it looks like that a collision only occured once in the history, deliberately engineered
15:42 przemek joined #metacpan
15:43 daxim what breaks if you'd provide a permalink patterned like e.g. <http://cpan.metacpan.org/r​eleases/Foo-Bar-ver.tar.gz>?
15:43 oalders ether: Carwyn Ellis actually did all the work. i just merged it :)
15:44 oalders and, i think, broke the build
15:45 haarg daxim: only once in what history?
15:45 daxim of cpan
15:46 haarg what
15:46 haarg where did you see that?
15:47 daxim https://github.com/CPAN-AP​I/metacpan-web/issues/797
15:47 dipsy [ "latest release" link points to unrelated distribution · Issue #797 · CPAN-API/metacpan-web · GitHub ]
15:48 haarg so you saw one example ans assumed it was only one?
15:48 haarg DBIx-Class for example has duplicate release filenames on metacpan
15:49 daxim which version? how did that happen?
15:49 przemek left #metacpan
15:50 haarg 0.08209, which i made by accident because of a builder bug
15:51 haarg for anything other than the latest release, there real record of "authorized" releases.  anyone could release anything with a duplicate name, and it would pause would accept it just like any other file.  it just wouldn't get indexed without permissions.
15:52 daxim I want to understand how big the problem is, then
15:52 haarg what i'm getting at is that this isn't a metacpan problem.  this is a pause problem.
15:52 daxim if it fails in 1‰ of the cases, it's still a win for the packager
15:53 haarg and anyone could break it at any point in time
15:54 daxim yeah yeah, now you're arguing from some unknowable future
15:56 oalders can we approach this from a module rather than a release url? i.e. give me the tarball which has the authorized version of module X version Y?
15:58 daxim this works for dependency resolution, e.g. Requires and BuildRequires, but Source wants a URL
15:59 daxim of the release tarball
15:59 oalders right. i'm saying provide a url, but give it a module name rather than a release
15:59 oalders and the endpoint can handle the logic from there
16:01 ether hey guys, just skimming the last hour or so of activity -- if a new version of a dist breaks a downstream dependency, you can indicate that in metadata (and in the Makefile.PL/Build.PL) with Dist::Zilla::Plugin::Conflicts - so when a user installs that version they are told they also need to upgrade their other thing too
16:02 oalders nice
16:04 daxim oalders, worth a try, I want to experiment with that
16:04 amrita_ joined #metacpan
16:04 daxim most of the time it's clear what the name of the main module is
16:05 daxim so, how do we go forward from here concerning issues/835?
16:05 haarg daxim: you might look at cpanm's code.  it does Module::Name@version -> tarball resolution.
16:07 haarg imo 835 is a non-starter until pause starts caring about filenames
16:08 haarg but providing an endpoint that could give you the tarball for a Module+Version might be workable.  i'm not sure.
16:09 oalders daxim: if you want to note the module+version proposal in the ticket, that's a start
16:09 oalders then we just need a volunteer ;)
16:29 daxim added a note
16:33 Amrita joined #metacpan
16:40 amrita__ joined #metacpan
16:51 dolmen left #metacpan
17:07 Amrita_ joined #metacpan
17:15 Amrita joined #metacpan
17:19 oalders daxim++
17:20 ether daxim: well volunteered! :)
17:20 oalders :D
17:21 Amrita_ joined #metacpan
17:37 oalders lucs: did you want to send a pull request?
17:39 lucs I'm not sure my solution is ideal, it just fixes the test.
17:40 oalders well, if the tests relies on the order of keys, then that's a problem
17:40 lucs I'm guessing it doesn't really matter in which order the modules are placed in the 'provides' array, but I don't really know.
17:41 lucs oalders: Yeah, the tests do, and that's a problem, but does the code somewhere else do too?
17:41 oalders lucs: i don't know about the code, but essentially the test is flawed, so i'd say you could send that as a pull request
17:42 oalders provides can be sorted elsewhere if that's needed. i don't think the code expects it to be in any order
17:42 oalders (i haven't actually look at the test)
17:42 oalders *looked
17:42 lucs My point is, fixing the test will allow the test to pass, but maybe someone who understands the code better should vouch for the fact that the order doesn't really matter.
17:43 oalders lucs: sure, but that's what comments on pull requests are for :)
17:44 Trelane random question: has anyone ever written a custom facet for elastic search?
17:46 oalders Trelane: there are some examples of facets on the wiki. do those help?
17:46 lucs oalders: Okay, I'll prepare a pull request :)
17:46 oalders lucs++
17:46 Amrita_ oalders: Hi, just sent in my application for OPW :)
17:47 oalders Amrita: excellent :) should we find something for you to do?
17:47 Trelane oalders: which wiki?
17:47 Amrita_ yes and yes! :)
17:47 oalders Trelane: https://github.com/CPAN-API/​cpan-api/wiki/Beta-API-docs
17:47 dipsy [ Beta API docs · CPAN-API/cpan-api Wiki · GitHub ]
17:48 Trelane oalders: you misunderstand
17:48 Trelane I mean has anyone ever written a facet from scratch?
17:48 Trelane i.e. a new java class for aggregation
17:48 Trelane or is talk about java verboten ;-)
17:49 rwstauner Trelane: i don't believe there's been any java work done within metacpan yet
17:49 Trelane oh goody
17:49 rwstauner so have fun ;-)
17:49 Trelane FSVO fun, I will ;-)
17:50 oalders Trelane: apparently i did misunderstand :)
17:50 oalders Amrita: https://github.com/CPAN-AP​I/metacpan-web/issues/596
17:50 dipsy [ File and Dist diff is not obvious · Issue #596 · CPAN-API/metacpan-web · GitHub ]
17:50 Trelane OPW?
17:50 dipsy OPW is http://perloasis.info - Orlando Perl Workshop, aka Perl Oasis
17:51 oalders that's just a UI fix, but you'll need to get metcpan-web up and running to fix it. that doesn't take very long
17:51 oalders we can help you with any issues you come across
17:51 Amrita joined #metacpan
17:52 * Trelane decides to write a perl prototype first
17:54 oalders Amrita: i posted an issue for you a few minutes ago. not sure if you saw it before your client disconnected
17:56 przemek joined #metacpan
17:58 Amrita_ oalders: thank you, I did get the link. Going through it now- i'll try and get metacpan-web up and running now then..
17:59 oalders Amrita++
18:01 rwstauner welcome aboard Amrita++
18:02 mst please please please can we call it OPfW
18:02 mst OPW is a conference.
18:03 Amrita_ rwstauner: thank you so much :]
18:04 rwstauner thanks for your help
18:05 oalders i guess that is confusing, but i'm not sure it's an acronym that will be bandied about a lot
18:06 mst it already caused mass confusion on #dancer
18:06 oalders heh
18:06 mst and dipsy kinda demonstrated the problem here already
18:06 oalders i accept your proposal.
18:07 oalders OPfW it is
18:07 mst cheers.
18:13 Trelane okay, what's OPfW?
18:13 Trelane ;-)
18:13 mst outreach program for women
18:14 Trelane ah ha.  And there I was thinking it might be ottawa perl workshop or somesuch
18:17 oalders https://live.gnome.org/OutreachProgramForWomen
18:17 dipsy [ OutreachProgramForWomen - GNOME Live! ]
18:25 amrita__ joined #metacpan
18:29 trs also http://news.perlfoundation.org/opw/
18:29 dipsy [ The Perl Foundation: OPW Archives ]
18:33 przemek joined #metacpan
18:37 Mike-PerlRecruiter_ joined #metacpan
19:02 Amrita joined #metacpan
19:35 amrita__ joined #metacpan
20:03 przemek joined #metacpan
20:05 dpetrov_ joined #metacpan
20:09 Amrita joined #metacpan
20:38 tianon joined #metacpan
20:44 amrita__ joined #metacpan
21:21 Amrita joined #metacpan
21:35 przemek joined #metacpan
21:58 Amrita joined #metacpan
22:31 amrita_ joined #metacpan
23:04 Amrita joined #metacpan
23:18 ether on https://metacpan.org/module/CURTIS/WWW-​KrispyKreme-0.01/lib/WWW/KrispyKreme.pm -- there is a 'go to latest' link that goes nowhere
23:18 dipsy [ WWW::KrispyKreme - root namespace for Krisy Kreme modules - metacpan.org ]
23:18 ether this 0.01 release was *just* uploaded
23:18 ether so is the heuristic for including that link incorrect?
23:19 ether perhaps "the release link for this dist does not map to this release" (which it wouldn't, if it's not indexed yet)?
23:36 thaljef joined #metacpan
23:41 Amrita joined #metacpan
23:49 mo ether: give it a few minutes

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