Perl 6 - the future is here, just unevenly distributed

IRC log for #metacpan, 2015-09-17

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

All times shown according to UTC.

Time Nick Message
01:22 haarg they are already going with META6.json
01:23 haarg i'm just wondering if it would be better to separate the files more
05:22 melo joined #metacpan
06:29 melo joined #metacpan
08:05 neilb joined #metacpan
09:55 melo joined #metacpan
10:08 margeas joined #metacpan
10:13 ilmari perl6 dists also have to be uploaded to a subdirectory matching /^Perl6\b/
11:18 DrHyde joined #metacpan
11:18 DrHyde hello gang
11:19 DrHyde ranguard set up accounts for me on {bm,lw}-mc-03.metacpan.org but they seem to be rejecting my ssh key
11:19 DrHyde are the perms right on my .ssh dir and contents?
11:19 DrHyde extra whitespace snuck in to authorized_keys via cut n paste?
11:29 jdv79 sorry - had to duck out y'day.  as ilmari said they are in $userhome/Perl6/ on cpan and skip the p5 indexer if it has a META6.json
11:30 jdv79 refinements may be needed but seems good enough - for now:)
11:31 haarg DrHyde: can you post your key so i can verify?
11:32 haarg username is davidc
11:32 jdv79 in other words, https://github.com/andk/pause/blob/master/lib/pause_1999/edit.pm#L1613-L1617
11:34 DrHyde ah, i was trying username david
11:36 DrHyde and if i try with the right username it all works.
11:36 DrHyde assumptions--
12:12 alexey joined #metacpan
12:27 vroom joined #metacpan
12:34 alexey Hi folks. I'd like to contribute to metacpan , who could I talk with about it?
12:41 jdv79 github?
12:41 jdv79 ;)
12:43 alexey yeah I was there if you metapcpan github pages
12:43 alexey I'd like to suggest some monitoring contribution, but not sure if metacpan really needs this
12:44 jdv79 afaik ranguard is the sa-ish and its only 1 box maybe 
12:44 jdv79 well, alrighty then 
12:44 alexey joined #metacpan
12:45 jdv79 afaik ranguard is the sa-ish and its only 1 box. not sure .
12:45 ranguard alh: hi
12:45 ranguard alexey: hi even (sorry alh!)
12:45 alexey hi!
12:47 ranguard https://github.com/CPAN-API/ has all our configs, including https://github.com/CPAN-API/metacpan-developer ( which sets you up a VM to use for developing ) and lots of other useful repos
12:48 alexey I wonder if metacpan needs some automatic smoke tests or monitoring?
12:49 ranguard we use Travis for testing ( https://github.com/CPAN-API/cpan-api/ and https://github.com/CPAN-API/metacpan-web see .travis.yml)
12:49 alexey or smoke test suite for developers?
12:49 ranguard alexey: improving our tests them selves would be great!
12:50 ranguard we are half way through a project to upgrade our Elasticsearch backend, and one of the issues is how to compare the current with the new
12:51 jdv79 testing with massive fixtures - always a joy
12:51 ranguard I can talk about infrastructure and setups, but for the actual code you'd want to discuss with..
12:51 alexey <ranguard> I have looked at some code base like - https://github.com/CPAN-API/metacpan-web/blob/master/lib/MetaCPAN/Web/Test.pm and https://github.com/CPAN-API/metacpan-web/tree/master/t ... But what I'd like to suggest is not a unit test style, but kind of sanity checks with curl against a real servers
12:51 ranguard oalders: ^^ :)
12:52 ranguard rwstauner: ^^
12:52 jdv79 alexey: and?
12:53 alexey jdv79: I need to know if it meets a real needs of metacpan ...
12:54 jdv79 could you clarify that? what is "it"?
12:54 ranguard alexey: mc-web is just a wrapper to cpan-api (api.metacpan.org) - so it has a 'real' server for other tests, but I'm probably missing the point, oalders usually wakes up in a couple of hours :)
12:54 jdv79 testing anything of any real complexity typically doesn't happen "purely"
12:55 alexey ranguard:, jdv79: ok. I mean if metacpan sites/services needs a web appliaction/services smoke tests at all, that is it
12:55 jdv79 unit, integration, regression, etc...  testing are not mutually exclusive:)
12:56 jdv79 still a bit vague.  do you mean a harness/runner setup or a test suite?
12:59 alexey jdv79:  take a look at this - http://swatpm.org/search?search_query= I think that swat might be helful to test metacpan services in some way, but first I need to know if there reals needs and if they might be covered by what I'd like to suggest ( swat )
13:00 alexey jdv79: this is a information on swat - https://github.com/melezhik/swat
13:04 jdv79 so its a framework,  and a runner/report service, for web testing?
13:05 jdv79 oh, no - its just a search site
13:05 jdv79 forgive me if i'm wrong, but i see little to noe value there
13:05 jdv79 *no
13:06 alexey jdv79: swat is a DSL/framework to  develop/run web application smoke tests; swatpm.org currently is just a "showcase" for existed swat packages -  a portable swat tests ( run by swat )
13:06 DrHyde left #metacpan
13:07 alexey jdv79: my suggestion to create some  tests for various metacpan services ( any ) using swat
13:08 alh ranguard: So no hi for me? :(
13:08 ranguard alh: hi, feel free to send patches :)
13:08 alh Hahahahah, well played
13:08 alh Hi :)
13:10 jdv79 my first reaction is web teting with the existing cpan tools (mech/cat helpers/etc) are already pretty decent.
13:11 jdv79 and the only thing that would be a killer feature of a newcomer would be real js/viewport testing
13:11 alexey jdk79: "js/viewport testing" what do you mean?
13:11 jdv79 imho of course
13:12 jdv79 i've been building and testing web type stuff mostly with perl for ~15 years - for context
13:13 jdv79 as in - a webpage has some js which is crucial to functionality - it needs to actually run in a "browser" (headless would be best) in order to really test the thing
13:13 alexey jdk79: ok, I see
13:13 jdv79 phantomjs seemed promising last i looked but never got a chance to give it a real go
13:14 alexey jdv79: not for js testing of course ; it based on curl
13:15 jdv79 yeah, which is just like the rest of already existing tools
13:15 alexey jdv79: I know that. I guess swat have some advantages even though
13:16 jdv79 i in no way speak for metacpan.  maybe they would disagree.
13:17 jdv79 what are those advantages?  the readme looks like just another web testing tool
13:19 alexey jdv79: 1) swat packages are portable test suites, distributed as cpan packages 2)  a simple DSL to make test developments process very fast and quite easy
13:21 alexey jdv79: idea is to create a test suite and let other to use it right away; Often tests are coupled with application,  it is ok , but sometimes it is useful to decouple test from application
13:23 margeas joined #metacpan
13:24 vroom joined #metacpan
13:28 jdv79 why is having a seperate test suite pkg useful?
13:29 alexey jdv79: simple DSL does not mean restricted , but generally you even do not need to be coder to start writing a swat test. Then of course you might to bring more complexity , that is where perl  knowledge is required.  swat might goes both ways of usage
13:29 alexey jdv79: good question to ask ... just a second
13:32 alexey jdv79: there are some areas when you need to share a test suite with others because likely application test suite is running again is installed on many servers and environments,  but when it comes to web monitoring people do need to a have an application ( in source code or what ever ) at all to test / monitor it
13:33 alexey application( a test suite is running against ) is installed on many servers and environments
13:33 ranguard jdv79: we wrote https://metacpan.org/pod/Test::WWW::Selenium::More at work for JS testing
13:33 alexey like people testing/monitoring perl appliaction or ruby application or whatever language application might do not know perl or ruby language at all
13:34 alexey but they need a way to test
13:34 jdv79 not to belabour my point but I think BDD is a bad idea in general.  but i've been burnt by it before and wrote a BDDD frame once.  idk.  you'd have to justify it more i think.  prototype it and show why its better
13:34 jdv79 good luck:)
13:35 jdv79 last time i tried to use selenium (2 years ago) it broke horribly
13:35 ranguard https://metacpan.org/pod/distribution/Test-BDD-Cucumber/lib/Test/BDD/Cucumber/Manual/Tutorial.pod
13:35 jdv79 ssl switching, modals, basic auth - from memory
13:35 jdv79 I know of cucumber.  maybe i'm just an embittered relic...
13:36 * ranguard doesn't like cucumber at all
13:36 ranguard jdv79: we don't use any of those thing, but wrote that plugin to make it Moosely (reusable bits with chaining) and fix some ajax timing issues
13:36 jdv79 its yet another level of abstraction to extend/maint/debug.  and on top of it you have non-tech people being targetted authors so that's great.
13:37 jdv79 bleh
13:37 ranguard jdv79: yea, my thoughts exactly
13:37 ranguard I can get designers (with a small clue) to write tests like this: https://metacpan.org/pod/Test::WWW::Selenium::More#SYNOPSIS
13:38 ranguard and we then subclass it and have methods like `login_as_fresh_user()` which they can use and hide a dozen steps
13:38 jdv79 alexey: yeah, that was what selenium was "supposed" to make happen.
13:48 alexey ok, guys, if you feel there is no need for metacpan, it's ok,  the initial purpose was not to compare existed test tools/frameworks, but I understand that it may matter indeed. If metacpan need some testing - I'd glad to do it with swat, the documentation on what you'd gain is quite clear,  but if you still have some questions and concerns I'll gladly answer
13:49 jdv79 if something could be like selenium but on an embedded headless browser - i think that may have worth.
13:49 ranguard alexey: please hang around, oalders and rwstauner will have a better perspective :)
13:49 alexey Sorry I mean -  if you feel that there is no need in web testing  for metacpan ...
13:49 jdv79 esp if it embedded more than one browser - phantom is just webkit i thiink
13:50 ranguard by 'better' I mean, they have a clue about the testing, jdv79 and I don't with respect to metacpan, just our personal opinions
13:50 alexey ranguard: sure
13:54 jdv79 if i had spare time and was fired up about testing that's like what i'd attempt to build/have built.
13:55 jdv79 *likely
13:58 alexey jdv79: Indeed I used to play alot with cucumber/webdriver/capybara/headless browsers in a past, but now my interests are a rest api / web services smoke/sanity/(may be a bit more complex ) checks ...
14:10 jdv79 cool
14:11 alexey yeah that is why I probably might know the difference b/w many "sorts" of testing :)
14:12 jdv79 nice
14:37 jdv79 so, why did my p6 dist knock my p5 dist out of the authorized results?
14:37 jdv79 that's not cool man
14:38 alexey jdv79: sorry , if this for me?
14:40 jdv79 n.  i just noticed that what i thought was just a nuissance side effect also has a real bug
14:40 jdv79 in metacpan
14:40 alexey ahh, ok
14:41 jdv79 oalders: maybe a pointer on this one?
14:41 oalders just reading backlog
14:42 jdv79 i thought if the latest is "unauthorized" it would not cover up an older authorized.  but in this case maybe the p6 dist is semi-authorized:(
14:42 alexey jdv79:  ranguard: oalders: I am going to create an issue on metacpan github regarded my contribution suggestion with  expanded explanation, and some quotes from our IRC talk, is it ok?
14:43 jdv79 alexey: cool
14:44 oalders alexey: yes, thanks.  i checked out swat yesterday.  looks like a helpful project
14:44 haarg there also are multiple releases with the same author and file name (not full path) and metacpan is definitely getting confused there
14:45 haarg see this file listing https://metacpan.org/source/JDV/Data-Selector-1.01
14:45 jdv79 oh, sweet
14:45 oalders alexey: we have decent coverage in our test suite, but outside of that we don't have anything that actively hits endpoints looking for expected results
14:46 oalders we have some basic monitoring via panopta for CPU, disk space etc.  could possibly see if we can have that trigger some other scripts (swat?) and then get monitoring reports that way.
14:47 jdv79 i guess a p5 instance change to not touch p6 dists would fix the whole lot of these types of issues
14:48 haarg jdv79: not quite
14:48 oalders alexey: so, please do write this up as an issue on github :)
14:48 jdv79 haarg: do tell
14:48 haarg ignoring the p6 dist would fix some of these direct issues
14:48 haarg but not the general issue of the confusion with multiple files with the same name
14:49 jdv79 well, have to start somewhere.
14:49 haarg or a file of the same dist name bumping off the authorized (02packages) release
14:49 haarg well, the issues are mostly unrelated
14:49 haarg ignoring p6 dists is something we'll want too
14:50 jdv79 the p5 index does not mix in p6 dists
14:50 haarg metacpan doesn't differentiate though, which we'll want it to
14:51 jdv79 i was going to throw a global flag for instance type - p5 or p6 and then branch off that
14:51 jdv79 that should handle the major issues like this
14:52 alexey https://github.com/CPAN-API/metacpan-web/issues/1591 , probably was better  to move it to https://github.com/CPAN-API/metacpan-developer ?
14:52 haarg as i'm saying though - the issue you pointed out today has nothing to do with p6
14:54 jdv79 I missed that part.  so if i push an authorized p5 Foo-1.00 and then push an unauthorized p5 Foo-1.01 the Foo-1.00 will disappear from metacpan search results?
14:56 haarg i think it requires an authorized dist, but something like that
14:58 haarg also that the file browser mixes the files between multiple dists with the same author+filename (different dir)
14:59 jdv79 that's impossible with p5 though since dist/ver are unique
15:00 jdv79 the browser mixing i mean
15:00 oalders alexey++
15:01 jdv79 as for the other issue - that is weird.
15:06 alexey oalders: leave the issue at CPAN-API/metacpan-web/issues , in case you need , please move it to proper place ...thanks
15:06 oalders alexey: it's good where it is :)
15:06 oalders thanks!
15:08 alexey oalders: thanks. good
15:28 haarg jdv79: file path is unique, but afaik it will let you upload multiple files with the same file name
16:12 vroom1 joined #metacpan
16:23 neilb joined #metacpan
16:45 vanstyn joined #metacpan
17:08 jdv79 what?  pause will not by virtue of the uris table and its indices
17:22 haarg jdv79: PATH vs FILENAME
17:24 neilb joined #metacpan
17:36 jdv79 oh.  depends what you mean.  there is a basename user p5/p6 index.
18:11 Peter joined #metacpan
19:33 Ptolemarch joined #metacpan
19:51 jnbek joined #metacpan
20:49 margeas joined #metacpan
21:06 neilb_ joined #metacpan
21:12 Peter joined #metacpan
21:14 haarg so for my PR to do better in-dist pod linking, i'm basically going to be including workarounds for issue #1506 and #1509
21:15 haarg would probably be better if the data in ES could tell me the information i want, but that will be harder for me to do
21:26 punter joined #metacpan
23:22 Ptolemarch joined #metacpan

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