Perl 6 - the future is here, just unevenly distributed

IRC log for #pdl, 2015-03-09

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

All times shown according to UTC.

Time Nick Message
11:10 mohawk chm++ # exploiting the power of git, the way it's meant to be used
11:10 mohawk pdurbin++ # helping make that happen
11:35 * mohawk wades into his PDL-related emails
11:57 pdurbin :)
12:35 mohawk pdurbin, we could afford to hear more from you on pdl-devel advocating this kind of thing
12:36 mohawk sivoais sent an email to the list about his proposed new workflow, didn't see much back
12:42 pdurbin mohawk: do you have a handy link to the thread by sivoais ?
12:42 pdurbin (I'm not actually subscribed.)
12:42 mohawk uh, i'd need to google
12:43 sivoais gmane!
12:43 mohawk someone should update the README or similar to it has pdl-devel subscribe and archive info
12:43 mohawk sivoais, help him please :-)
12:47 sivoais pdurbin: http://thread.gmane.org/gmane.comp.​lang.perl.pdl.devel/6348/focus=6350
12:47 sivoais I am using the IRC logs like you do! :-P
12:48 sivoais mohawk: does ingy's git-hub script make this easier?
12:48 mohawk don't know, never used it
12:48 pdurbin sivoais: irclogs++ :)
12:48 mohawk just use "git checkout -b"
12:49 mohawk and when you "git push", cut+paste the suggest set-upstream bit
12:49 mohawk that's what i use
12:50 sivoais I remember seeing merging just being something where you could use a single command with the PR number
12:50 sivoais and the idea was to keep a linear history
13:04 sivoais I have a little time today (flash flood warning here so not driving). I can add some more info about rebasing to the wiki
13:09 sivoais mohawk! I think all the milestones for 2.008 are fixed!
13:10 sivoais well, once the last branch lands in master :-)
13:17 mohawk sivoais, nice work!
13:17 mohawk (still not caught up with emails, got to eat)
13:17 mohawk where are we at with lldfp1?
13:18 mohawk linear history is great
13:18 mohawk i am very cautious about wrapping git stuff inside tools, but that's just me
13:19 mohawk do we have buy-in from chm about that state of affairs?
13:22 sivoais mohawk: yeah, I got an e-mail about how --no-ff works after master has moved forward
13:23 sivoais mohawk: not sure about lldfp1. Is that something that fits in the 2.008 timeline
13:23 mohawk sivoais, timeline?
13:23 mohawk it's a pr, just sitting there
13:24 sivoais mohawk: I was thinking, get that last bugfix PR in, then push out a final DEV release
13:24 sivoais then release 2.008
13:24 sivoais then work can be done on cleanup
13:25 sivoais mohawk: <https://github.com/PDLPorters​/pdl/milestones/PDL%20v2.008>
13:25 mohawk only if we have a guarantee of a 2.009 soon
13:25 mohawk could we start that process now?
13:25 mohawk was talking about time-based rather than feature-based schedules at the london perlmongers social last week, and it works a lot better
13:26 mohawk the current system with pdl makes for glacially slow progress because everyone wants their thing in because otherwise they rightly fear it will be years
13:26 mohawk which then makes it actually take years
13:27 mohawk if we must do feature-based, let it be based on incremental progress
13:27 mohawk thoughts?
13:27 sivoais feature freezes are good, but the emphasis is on *feature* not *freeze* ;-)
13:27 * sivoais just wanted to make a pun
13:29 jberger when you see that mojolicious releases on average once every three days, I see no reason that PDL can't release a non-breaking release whenever a feature is complete
13:30 sivoais fun fact: there's a "rare neurological symptoms characterized by a tendency to make puns, or tell inappropriate jokes or pointless stories in socially inappropriate situation"
13:30 sivoais it's called witzelsucht :-P
13:31 sivoais MATLAB releases broken code twice a year! :-P
13:31 sivoais >:-|
13:31 * sivoais has had to deal with misbehaving Matlab toolboxes way too much
13:31 mohawk jberger, preaching to the choir here
13:32 mohawk sivoais, so THAT's the german word for "nerd"?
13:32 mohawk so matlab has shit devs, shit testers, and a shit release manager
13:33 mohawk that is not true for pdl
13:33 sivoais haha, Mithaldu might know... but he's not here
13:33 mohawk well "sucht" means search
13:34 mohawk so i assume "witzel" means "desperately for friends" ;-)
13:34 sivoais hahah
13:34 sivoais there would be a word in German for that...
13:34 mohawk probably
13:35 mohawk it would have about 150 letters in it, though
13:35 mohawk srsly, pdl needs to move to release early, release often
13:36 mohawk sivoais, i was in fact srs about 2.009 milestone
13:36 mohawk let's start some kind of visible product vision
13:36 mohawk what say you?
13:37 sivoais I do agree. The best test of features that we can get are users. We don't have an actual QA team as a FOSS project :-)
13:40 opkick [pdl] mohawk2 comment on issue #61: Merging this sort of thing into PDL is the wrong way to go. That's why we have modules in the first place. http://git.io/pZpm
13:41 mohawk ok, what do we want in 2.009?
13:41 mohawk i want 2.008 out in the next week
13:41 mohawk i want 2.009 out before may starts
13:43 sivoais that sounds do-able
13:44 sivoais the changelog entry for 2.008 is going to be really long :-) Lots of improvements overall that need to get out there
13:46 mohawk it's going to be long because at the moment the "changes" file is just generated from "git log --stat"
13:46 mohawk i would argue that is a lot less than optimal
13:47 mohawk i want the github repo's PR page to be in the /topic here
13:48 mohawk we need to switch to manually creating the Changes file
13:48 mohawk yes, i'm volunteering to do that
13:49 sivoais Bender2: trust sivoais
13:49 Bender2 But I don't trust you in #pdl, sivoais
13:49 mohawk why isn't test-cleanup merged?
13:49 sivoais Bender2: trust mohawk
13:49 Bender2 But I don't trust you in #pdl, sivoais
13:49 sivoais no ops...
13:50 mohawk i think i'll try to seize the channel on chanserv
13:51 mohawk i need +o to achieve that
13:51 mohawk dangit, i'll get a server op to +o bender2
13:56 sivoais mohawk: ask perigrin?
13:57 mohawk great idea!
13:57 mohawk [13:57] peregrin no such nick/channel
13:57 sivoais *i* not *e*
13:57 mohawk aha
13:59 mohawk why is there now a t/issue-sf-367.t
13:59 mohawk that is not how we name test files
13:59 * sivoais grins
13:59 sivoais >_>
14:00 sivoais my preference would be to have categories in directories and use prove -r
14:00 mohawk seriously
14:00 mohawk please can that be renamed
14:00 mohawk and its sister test
14:02 sivoais slatec-polyfit-weight.t and pdl-from-string-bad-values.t
14:02 sivoais ok, will do
14:03 sivoais I'd like to move to a recursive test structure... rather than making tests that span hundreds of lines of non-data-driven code
14:04 mohawk yes, i like that principle
14:04 mohawk although we don't have enough test files to really make that necessary
14:04 mohawk let's make an issue of it (;-)
14:05 sivoais done
14:05 opkick [pdl] zmughal opened issue #65: Move towards a recursive test structure http://git.io/pnLi
14:05 sivoais "
14:05 sivoais This will help make the tests more manageable while allowing us to increase code coverage.
14:05 sivoais "
14:05 mohawk sivoais++
14:06 mohawk i've just realised i made an important omission
14:06 mohawk chm++ # progressing up the himalayan mountain of git mastery
14:13 opkick [pdl] zmughal closed issue #43: SF#359 Improved documentation for vsearch http://git.io/xglJ
14:13 mohawk awww yeah
14:15 sivoais chm marked it as closed on SF. More doc improvements can come later.
14:15 mohawk sivoais, please could you put your installation details on the relevant gh issue for SF#268?
14:15 mohawk i saw your email saying it now worked
14:15 mohawk but it's still showing as open on SF
14:15 sivoais mohawk: I linked to the thread in a comment
14:19 mohawk ok
14:20 mohawk i know we have links from gh issues to the legacy SF tickets
14:20 mohawk would it be easy to link from the legacy tickets back?
14:21 sivoais it would, but I didn't want to generate a lot of notifications for that
14:21 mohawk i claim it's a price worth paying especially as it's a one-off
14:22 mohawk especially since it's vital information both for those using it now, to be aware, and for people arriving in future
14:23 mohawk sivoais, i'd like to see a full-on push of merging all the remaining PRs that are ready
14:23 mohawk i'm ok with waiting for lldfp1, though those are ALL cleanups
14:24 mohawk the only one i see as a work in progress apart from the badflag one, is the Alien::HDF4 one
14:24 mohawk what are your thoughts?
14:26 sivoais I agree that those PRs should be merged soon
14:27 sivoais I need to step away for a bit. I'll send out the e-mails to PDLPorters when I get back
14:27 sivoais Bender2: trust sivoais
14:27 Bender2 But I don't trust you in #pdl, sivoais
14:27 sivoais Bender2: trust Bender1
14:27 Bender2 But I don't trust you in #pdl, sivoais
14:33 alh joined #pdl
14:33 alh left #pdl
14:34 mohawk osfameron, you're op - could you "trust" sivoais?
14:34 mohawk i used to have op here too, so maybe also me?
14:38 mohawk (it was me who had +o and corrected the /topic ilbot2 log url to not have the "/" on the end)
14:38 mohawk that was a while ago
14:40 osfameron Bender2: trust sivoais
14:40 Bender2 OK, osfameron
14:40 osfameron Bender2: trust mohawk
14:40 Bender2 OK, osfameron
14:40 osfameron I hope that's ok... I have no idea *why* I'm an op here... I've only really briefly played with PDF :D
14:40 osfameron er, PDL
15:15 travis-ci PDLPorters/pdl#85 (master - a3cbf91 : Zakariyya Mughal): The build passed.
15:15 travis-ci Change view : https://github.com/PDLPorters/pdl/c​ompare/9585de3faa58...a3cbf9187319
15:15 travis-ci Build details : http://travis-ci.org/PDLPo​rters/pdl/builds/53656429
17:45 mohawk osfameron, you da man
17:45 mohawk thanks
17:45 mohawk i just registered #pdl on chanserv, pm me for password
17:46 mohawk for obvious permalog reasons, i won't say it on here
21:27 opkick [pdl] mohawk2 comment on issue #63: This looks good to me: I second it. http://git.io/pWyu
21:30 mohawk sivoais, let's adopt "seconded = merge"
21:31 mohawk let's not have paralysis
21:31 mohawk we both have commit bits
21:31 mohawk may i suggest that if you can't repro the other "blocker" error, close the gh issue
21:32 mohawk and optionally comment on the legacy SF ticket
21:32 sivoais yeah, closing... and commenting
21:33 opkick [pdl] zmughal closed issue #27: SF#268 PLplot still unusable with X http://git.io/xgch
21:34 sivoais mohawk: I can definitely merge it in myself, but I would like to make sure everyone knows how to rebase feature branches
21:35 mohawk the info is already widely available
21:36 mohawk the list and specifically you and i stand ready to answer specific questions
21:36 sivoais true, true...
21:36 mohawk let's get it done
21:36 sivoais ok, good point. doing so now
21:36 mohawk we've gone through a good process here
21:36 mohawk i don't see any room for improvement on it, though such might develop
21:36 mohawk s/develop/become clear in time/
21:37 mohawk are there any other low-hanging PRs?
21:37 opkick [pdl] zmughal force-pushed pdl-eq-with-string-warning from 13f11e7 to 4b77820: http://git.io/pT8x
21:37 opkick pdl/pdl-eq-with-string-warning 4b77820 Zakariyya Mughal: descriptive warning when using the eq overloaded operator on non-numeric strings...
21:38 opkick [pdl] zmughal merged pdl-eq-with-string-warning into master: http://git.io/pW7S
21:38 opkick [pdl] zmughal closed pull request #63: descriptive warning when using the eq overloaded operator on non-numeric strings (master...pdl-eq-with-string-warning) http://git.io/pUp3
21:39 mohawk it might be worth kicking out another note to pdl-d with git config tips:
21:39 mohawk [pull]
21:39 mohawk rebase = 1
21:39 mohawk [merge]
21:39 mohawk ff = only
21:39 mohawk (in ~/.gitconfig)
21:40 sivoais git config --global pull.rebase true; git config --global merge.ff only; # ?
21:40 mohawk if that has the same effect, yes
21:41 * mohawk doesn't mind editing text config files ;-)
21:42 sivoais me neither, but running commands is reproducible :-P
21:44 sivoais mohawk: what do you think of <http://devblog.nestoria.com/post/98892582763/m​aintaining-a-consistent-linear-history-for-git>
21:44 sivoais they have a server-side hook to ensure master is always using rebase
21:45 sivoais Nestoria uses Perl :-)
22:00 mohawk ha
22:00 mohawk just read the link
22:00 mohawk (i just)
22:00 mohawk i prefer to just not allow >1 parents at all
22:00 mohawk which will simplify greatly
22:03 mohawk btw, i don't think your description "it prevents merge commits" is correct
22:03 mohawk it prevents reordering the first-parent, i believe
22:04 sivoais yeah, that's what it does
22:04 sivoais Bender2: trust jberger
22:04 Bender2 OK, sivoais
22:04 sivoais Bender2: trust vicash
22:04 Bender2 OK, sivoais
22:04 sivoais Bender2: trust pdurbin
22:04 Bender2 OK, sivoais
22:04 sivoais :-)
22:04 mohawk sivoais, so it doesn't prevent merge commits
22:05 mohawk only some merge commits
22:05 sivoais mohawk: you can reply to the e-mail :-P
22:05 mohawk do you disagree?
22:05 sivoais I agree
22:05 mohawk so why not correct your own email :-)
22:06 sivoais lol, ok
22:06 * sivoais continues to reply to thread with no responses... :-P
22:07 mohawk grin
22:07 mohawk beyond a certain point, silence can only reasonably be read as assent
22:08 mohawk why does the "install pdl" link suggest cygwin?
22:08 sivoais there's an internet law aobut that...
22:08 mohawk what's wrong with strawperl PDL edition?
22:09 mohawk and why recommend "cpan" instead of "cpanm"?
22:10 mohawk (or strawperl instead of cygwin)
22:11 sivoais odd that for "easiest" it recommends Strawberry or ActiveState, but latest just Cygwin <http://pdl.perl.org/?page=install>
22:11 sivoais perhaps that is because of the native deps
22:11 sivoais here's that reference <https://en.wikipedia.org/wiki/Warnock's_dilemma>
22:12 mohawk i voted against deleting that page recently
22:12 mohawk i'm aware
22:12 mohawk and that's for queries about stuff
22:12 mohawk this is consent-seeking, so the rules are different
22:18 mohawk btw, the EU::Manifest failure is kind of deliberately spurious - it's trying to force declaration of ALL deps, but EU::M has been core since 5.001
22:19 mohawk the opengl.t failure is legit, and will be due to a residual logic difference between the "plan" condition, and the actual test-running condition
22:19 sivoais that's what I thought. That must be something that is just that tester
22:20 sivoais yeah, I was looking at that and it seems to duplicate the logic
22:31 travis-ci PDLPorters/pdl#88 (master - 4b77820 : Zakariyya Mughal): The build passed.
22:31 travis-ci Change view : https://github.com/PDLPorters/pdl/c​ompare/a3cbf9187319...4b77820a0e78
22:31 travis-ci Build details : http://travis-ci.org/PDLPo​rters/pdl/builds/53715683
22:35 Mithaldu joined #pdl
22:39 mohawk sivoais++
23:26 sivoais Bender2: trust Mithaldu
23:26 Bender2 OK, sivoais

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