Perl 6 - the future is here, just unevenly distributed

IRC log for #inline, 2015-02-20

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

All times shown according to UTC.

Time Nick Message
09:07 rurban1 joined #inline
09:40 rurban2 joined #inline
09:41 rurban2 left #inline
12:09 Mithaldu ingy: makestub doesn't write files with binary
12:43 rurban1 left #inline
12:46 Mithaldu ingy: some issues when trying to dzil it with my own bundle, thrown up by the xt tests: the generated ::Inline module has no pod or version, suggestions how i'd get those in there?
12:50 Mithaldu ingy: i'm also seeing warnings because i'm doing ::Inline->import( C => <<'...' ); multiple times in a file
12:50 Mithaldu Subroutine dl_load_flags redefined at (eval 22) line 8.
12:50 Mithaldu Subroutine Acme::MITHALDU::XSGrabBag::mix redefined at (eval 22) line 9.
12:50 Mithaldu mix is the first function declared
12:50 Mithaldu maybe import should die when being run multiple times from the same caller?
13:02 Mithaldu ingy: and released: https://metacpan.org/release/MITHALDU/Acme-MITHALDU-XSGrabBag-1.150510
13:28 Mithaldu and a fix release already because i'm dumb: https://metacpan.org/release/MITHALDU/Acme-MITHALDU-XSGrabBag-1.150511
13:45 Mithaldu ingy: thanks to you too :) https://twitter.com/mithaldu/status/568767715781058561
14:29 ingy Mithaldu: I'll add your dist to the Inline::Module devel (xt) tests
14:29 Mithaldu cheers
14:30 ingy The I'll test for any actions that are different or causing problems that you mentioned
14:30 ingy *Then
14:30 Mithaldu \o/
14:31 Mithaldu i just hope my feedback is useful :)
14:39 mohawk Mithaldu, the canonical way to do that is with "bind" not "import"
14:40 Mithaldu mohawk: doesn't seem to be documented here: https://metacpan.org/pod/distribution/Inline-Module/lib/Inline/Module.pod
14:42 Mithaldu "Although bind() is a powerful feature, it is not recommended for use in Inline based modules. In fact, it won't work at all for installable modules."
14:42 Mithaldu in Inline.pm
14:42 Mithaldu of note:
14:42 Mithaldu my issue here is that i want to have pod/function/pod/function
15:01 mohawk the next great frontier for I::M is to have multiple code snippets from the same file "just work"
15:01 mohawk it would have to bundle them up in one though, possibly with an INIT (?) block
15:03 GitHub135 [inline-module-pm] ingydotnet created PR/31 (+8 new commits): http://git.io/AcUU
15:03 GitHub135 inline-module-pm/PR/31 5ff55da Ed J: Make fixblib work
15:03 GitHub135 inline-module-pm/PR/31 aa4206f Ed J: Separate target build_inline aids testing
15:03 GitHub135 inline-module-pm/PR/31 e30eb09 Ed J: Use File::Copy not link for SO
15:04 ingy Mithaldu: can you try something for me
15:04 ingy pull I:M
15:04 Mithaldu ingy: sure, what?
15:04 ingy then run: prove -lv test/devel/test-inline-modules.t :: a-m-x
15:04 Mithaldu sec
15:05 ingy mohawk: also, I added stuff to show errors and help get prereqs
15:05 ingy you might want to try it and see if it's ok
15:05 travis-ci ingydotnet/inline-module-pm#84 (PR/31 - 17416a5 : Ingy döt Net): The build passed.
15:05 travis-ci Change view : https://github.com/ingydotnet/inline-module-pm/compare/5ff55dac6726^...17416a549201
15:05 travis-ci Build details : http://travis-ci.org/ingydotnet/inline-module-pm/builds/51520909
15:05 ingy multiple inline module sections should just work now
15:05 ingy if not make an issue
15:06 Mithaldu ingy: ext/test-more-bash/ext/bashplus/test/source-bash+-:std.t
15:06 Mithaldu that file is invalid on windows
15:06 ingy oh windows :)
15:06 Mithaldu D:\cpan\inline-module-pm>prove -lv test/devel/test-inline-modules.t :: a-m-x
15:06 Mithaldu test/devel/test-inline-modules.t .. Can't exec /usr/bin/env at test/devel/test-inline-modules.t line 1.
15:06 ingy yeah ok, nm.
15:07 ingy mohawk: if you run: CLONE_OK=1 prove -lv test/devel
15:07 ingy it should work now
15:07 ingy if you leave off the CLONE_OK and the repo is missing it will tell you what you can do
15:08 ingy I want the bash stuff working on windows. I've heard that it's not far off
15:08 ingy but it's not high prio today
15:32 mohawk i wish you would use perl not bash
16:05 mohawk it does seem to make things unnecessarily difficult
16:31 mohawk ingy, i cherry-picked your commit over to my fixblib branch
16:31 mohawk ran CLONE_OK=1 prove -lv test/devel
16:31 mohawk got slightly further:
16:31 mohawk # Testing 'acme-math-xs-pm' branch 'cpp'
16:31 mohawk Bailout called.  Further testing stopped:  Bailing out on status=1
16:31 mohawk what now?
17:03 ingy mohawk: can you push that?
17:03 ingy to another branch if you want
17:04 ingy then I'll try what you have in a clean env
17:04 ingy I should get the same error
17:05 ingy I'll try that now with my HEAD
17:06 mohawk ingy, i pushed your extra commit on my PR branch
17:08 mohawk got it?
17:09 GitHub79 [inline-module-pm] ingydotnet pushed 1 new commit to master: http://git.io/ACI1
17:09 GitHub79 inline-module-pm/master 1419022 Ingy döt Net: Make tests better...
17:10 ingy dang it! I didn't push that to master :\
17:10 ingy cherry-picked and done now
17:11 mohawk wait, i'll rebase and force push
17:12 ingy mohawk: I got your error
17:12 ingy will fix it
17:12 mohawk repushed
17:12 mohawk ok
17:24 ingy the problem is annoyying but I think I have a fix
17:31 ingy gbr -a | cut -c3- | grep ^remotes/ | cut -d' ' -f1 | cut -d/ -f3- | grep -v HEAD | xargs -n1 git checkout -q
17:31 ingy I win!
17:43 mohawk and everyone who's on windows loses
17:43 mohawk nice work
17:59 ingy mohawk: what's not correct, and not appreciated. one person (Mithaldu) loses the ability to run dev only tests, and it's just a bug
17:59 ingy *that's
18:00 GitHub10 [inline-module-pm] ingydotnet pushed 1 new commit to master: http://git.io/ACg9
18:00 GitHub10 inline-module-pm/master 3641499 Ingy döt Net: Fixed tests to clone correctly
18:00 mohawk looks to me like your long, tortured pipeline above could have been a perl oneliner mapping over output from "gbr -a |"
18:00 ingy it's way easier for me to write system interaction tests with bash than with perl.
18:01 Mithaldu i don't mind if some tests get skipped if the whole thing still works
18:01 mst sometimes win32 is different enough that you have to say "this is currently broken on win32, we acknowledge that this is a bug, but we've got bigger fish to fry right now"
18:01 ingy mohawk: that should run for you now with CLONE_OK
18:01 mst especially when the final result still works correctly
18:02 ingy mohawk: it will fail on some prereq modules, but the errors should print out and be obvious
18:02 mohawk CLONE_OK=1 prove -lv test/devel
18:03 mohawk error: pathspec 'cpp' did not match any file(s) known to git.Bailout called.  Further testing stopped:  Bailing out on status=1
18:03 Mithaldu also
18:04 Mithaldu i'm of the opinion that if ingy can do things quickly with bash, then that's fine and it's good for him to do prototyping
18:04 Mithaldu and maybe in a year we can port it to perl when it's not being changed anymore
18:04 mohawk that's a personal choice for him, and that's great
18:05 mst for systems-y stuff it's a pretty reasonable choice
18:05 mohawk however, it makes it difficult for others to dev the module / run all the dev tests (which is part of proper dev)
18:05 mst the point about testing is absolutely valid
18:05 mst no. the fact that the code is new and still contains bugs makes it difficult
18:05 mst that happens with new code
18:05 ingy mohawk: you got that on inline-module-pm master?
18:05 mohawk and there's a point of non-agreement
18:05 mohawk ingy, no, on my branch
18:06 mohawk AND on my master
18:06 ingy you cherry-picked the commit from 3 mins ago
18:06 ingy ?
18:06 ingy 6 min
18:06 mohawk my master is up to date with yours
18:06 mst mohawk: your complaint is morally equivalent to people looking at perl code they don't understand and saying "this would be easier if we wrote it in python"
18:07 mst it comes across as incredibly condescending
18:07 ingy sec
18:07 mohawk mst, that's an interesting view
18:07 ingy mst: I agree, and I'm willing to fix anything and help with anything
18:08 mst mohawk: well, it's how your remarks are coming across to other people. if you meant something different, evidently you need to step back, apologise for the accidental offensiveness, and rephrase
18:08 ingy mohawk just happens to be the first one working with these tests from scratch, so bugs expected (and hopefully just fixed)
18:09 mohawk i apologise for being accidentally offensive
18:11 Mithaldu mohawk: fwiw, as much as i hate bash, because i don't understand it, i realize that causation and keep that hate to myself, in other words, even i'm agreeing with mst on that
18:13 mohawk i think a useful way to repro any other person trying to run these dev tests is to make a separate unix user, have them clone i-m-pm, then run the tests
18:13 mohawk rather than going via git and my hard disk
18:18 ingy mohawk: https://gist.github.com/fe77c9c39e861f0b778c
18:19 mohawk ingy, you may as well delete your repo's PR/30 since that's all done now
18:19 ingy ok
18:19 ingy (set -ex; rm -fr mohawk; mkdir mohawk; cd mohawk; git clone https://github.com/mohawk2/inline-module-pm; cd inline-module-pm; CLONE_OK=1 prove -lv test/devel)
18:19 ingy mohawk: That's running completely from scratch
18:20 ingy except I have the various prereqs
18:20 ingy CPAN prereqs
18:20 ingy using your master
18:21 mohawk on the command line, i go into the a-m-x-pm dir and do git checkout cpp, it works fine
18:22 mohawk ingy, did you do the above on a separate unix user?
18:22 GitHub96 [inline-module-pm] ingydotnet deleted PR/31 at 17416a5: http://git.io/ACXC
18:23 ingy I did it in my home dir
18:23 ingy it creates the new ./mohawk/ subdir
18:23 mohawk i gathered that part
18:24 mohawk the separate user would eliminate the possibility of finding stuff you already have cloned / correctly setup
18:24 ingy you can run it as any user you want
18:24 mohawk (in your current dir structure)
18:25 ingy it works by making a clone of the clone
18:25 mohawk so after having done "git checkout cpp; git checkout -" on my a-m-xs dir, now it's getting to test 4
18:25 ingy so not really a problem
18:25 mohawk my hypothesis is that your code searches both ".." and "../.." for any repos
18:25 mohawk so you not following my suggestion means you'd get a false positive on "it works ok"
18:26 mohawk since it's not reproducing the conditions of a completely new user arriving fresh
18:26 mohawk i'm sure you know best
18:26 ingy mohawk: can you run: (set -ex; rm -fr mohawk; mkdir mohawk; cd mohawk; git clone https://github.com/mohawk2/inline-module-pm; cd inline-module-pm; CLONE_OK=1 prove -lv test/devel)
18:26 ingy in /tmp/
18:27 ingy and it will guarantee to not get anything stale
18:27 ingy /tmp/ is not necessary, but might feel more safe :)
18:27 ingy just don't run in /home/ if your $USER == mohawk :)
18:28 ingy it expects the repos to be in ..
18:29 ingy we can change that if needed, but the repos don't really change much
18:29 ingy we are just checking that inline-module still works on them
18:29 ingy and also they can be on some other branch
18:30 ingy because the test clones the .. repo and changes to the branch it wants
18:30 mohawk just did your above command line
18:30 mohawk fails at test 4 again:
18:30 mohawk not ok 4 - Alt-Acme-Math-XS-CPP-0.0.5/MANIFEST exists
18:32 ingy not sure how that's possible, since everything is fresh
18:33 ingy anyway, I'm out of time for now…
18:34 mohawk i will place a small bet that if you actually try my suggestion it might add something
18:35 mohawk s/might/will/
18:35 ingy which suggestion?
18:36 mohawk i think a useful way to repro any other person trying to run these dev tests is to make a separate unix user, have them clone i-m-pm, then run the tests
18:36 mohawk from 23 mins ago
18:36 ingy I'll run this on some vm right now
18:43 ingy mohawk: I did get an weird error on the dzil branch
18:43 ingy but the cpp tests all passed ok
18:44 mohawk oh
18:44 ingy mohawk: also https://github.com/ingydotnet/inline-module-pm/blob/master/test/devel/test-inline-modules.t#L150
18:44 ingy for running the tests individually
18:44 ingy ok, I'll be back in the future
18:44 ingy ☺/
18:45 mohawk enjoy
20:51 rurban1 joined #inline

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