Perl 6 - the future is here, just unevenly distributed

IRC log for #inline, 2015-02-17

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

All times shown according to UTC.

Time Nick Message
00:42 davido_ltp joined #inline
02:48 ilbot2 joined #inline
02:48 Topic for #inline is now LOG: http://irclog.perlgeek.de/inline | TPF Grant Work in Progress | http://inline.ouistreet.com | Discussion of Inline.pm and ILSMs | https://github.com/ingydotnet/inline-pm
08:59 rurban1 joined #inline
09:25 Mithaldu i'll probably be wrapping the 96 bit mix function here in a module: http://web.archive.org/web/20071223173210/http://www.concentric.net/~Ttwang/tech/inthash.htm
09:25 Mithaldu anyone got suggestions for a name for that?
09:31 Mithaldu ugh, inline::c is failing on windows
09:36 Mithaldu why does the inline-c-pm readme not have instructions on how to build the thing from repo?
09:59 Mithaldu why does 27 fail the make test steps when run as perl -Ilib, but not under prove?!
11:44 Mithaldu hrm, the make install bit of 27 fails because it tries to create .packlist twice
12:09 Mithaldu fwiw, it seems like this is a bug in ExtUtils::Packlist
12:29 Mithaldu gotta revise that: the bug is the presence of a packlist in blib at all
12:32 Mithaldu what's this about anyhow?
12:32 Mithaldu File::Spec->catfile($autodir,'$modfname.bs')
12:33 rurban1 joined #inline
12:43 Mithaldu ingy: you guys optimized Inline to be as clever and impenetrable as possible, huh?
12:47 rurban1 left #inline
13:02 Mithaldu no wonder it doesn't find this:
13:02 Mithaldu D:\cpan\inline-c-pm\_Inline_27inline_maker.297628\src dir\$(INST_ARCHLIB)\auto\Math\Simple\.packlist
13:02 Mithaldu or this:
13:02 Mithaldu D:\cpan\inline-c-pm\_Inline_27inline_maker.297628\src dir\$(INST_ARCHLIB)\auto\Math\Simple\$modfname.bs
13:06 Mithaldu mohawk: this might be your specialty
13:06 Mithaldu $(PERL) -Mblib -MInline=NOISY,_INSTALL_ -MMath::Simple -e"Inline::satisfy_makefile_dep({API => {modinlname => 'Math-Simple.inl', module => 'Math::Simple'}});" 1.23 $(INST_ARCHLIB)
13:06 Mithaldu why is dmake passing $(INST_ARCHLIB) on to perl as it is instead of interpolating the variable?
13:07 Mithaldu the $(PERL) *does* get interpolated
13:20 Mithaldu ah, it's the {
13:21 Mithaldu on that note, let me express my most sarcastic joy over the prevalence of oneliners in makefiles
13:35 GitHub14 [inline-pm] wchristian opened pull request #51: avoid {}s in generated Makefile code to avoid breaking dmake (master...patch-1) http://git.io/NN2R
13:41 GitHub83 [inline-c-pm] wchristian comment on issue #36: I just made a pull request to Inline.pm that should fix the root issue for this. A fix for this issue then would be a version dependency bump.... http://git.io/NNVi
13:52 ingy thanks Mithaldu
13:52 Mithaldu ingy!
13:52 Mithaldu you're there, yay
13:52 Mithaldu i got questions!
13:52 ingy mohawk has answers!
13:52 ingy ;)
13:52 Mithaldu do you think the Inline::C compilation always generates a packlist file?
13:53 ingy what's a packlist file?
13:53 Mithaldu also, do you think $autodit should always exist?
13:53 Mithaldu *$autodir
13:53 ingy hmm, I'll have to dig as much as you
13:53 Mithaldu that being the directory in which the cleanup phase looks for compilation artifacts like packlist, bs, exp, lib
13:53 ingy maybe davido___ knows
13:54 ingy Mithaldu: define always
13:54 ingy brb
13:54 Mithaldu also, a packlist file is a thing ExtUtils::Install generates to log which files were installed
13:54 Mithaldu always: whenever Inline::MakeMaker and Inline::C are used together
14:00 ingy Mithaldu: the only thing I think is that Inline should make people happy. Everything else is just bugs and fixes. So let's back up. What's making you unhappy wrt Inline?
14:01 ingy also you are contending with my early morning routine so I'll be a bit laggy…
14:01 Mithaldu mainly that the cleanup is currently completely buggered due to reasons for which i am providing commits right now
14:01 ingy what's the cleanup?
14:02 Mithaldu also on a meta level that it's awfully hard to debug because it has so many modules that do more than just exporting functions when they're use'd
14:02 ingy basically I have no idea yet, of your mental model / state of affairs
14:02 Mithaldu sub Inline::C::celanup
14:02 ingy but we can resolve that
14:02 Mithaldu sub Inline::C::cleanup
14:03 Mithaldu mind, due to the extremely hard debuggerability i'm mainly guessing at what i am seeing, but:
14:03 ingy ok. realize that much of this code is 14.5 years old, and I've done a few other things since then so it might take me a few minutes to sync up
14:04 Mithaldu in order to `make` an inline c module into XS code, Inline::C uses Inline::MakeMaker to do a fake install into blib/arch/auto
14:04 ingy do you have an open project you are trying Inline (on windows I assume) on?
14:05 Mithaldu since that generates compilation artifacts in blib that are not supposed to be installed into site/lib, Inline::C tries to delete some of them in the function i mentioned
14:05 ingy ok
14:05 Mithaldu however due to I::MakeMaker generating makefile code that's incompatible with dmake it never passed the right arguments so the cleanup phase could actually clean up
14:06 Mithaldu and the cleanup phase didn't try to verify what it's doing, so it simply failed to delete anything, silently
14:06 Mithaldu which caused the test 27 in Inline::C to fail when trying to install on windows
14:07 Mithaldu (and i found an additional thing about the cleanup sub that seems to be a general bug)
14:07 Mithaldu > do you have an open project you are trying Inline (on windows I assume) on?
14:07 Mithaldu i'm planning to make an inline c module, described in this channel 4.5 hours ago
14:08 Mithaldu but that got derailed by Inline::C failing to install
14:08 Mithaldu ingy: hope that summarizes the situation a bit :)
14:11 GitHub193 [inline-c-pm] wchristian opened pull request #37: make cleanup() more robust against bogus input and unexpected file system layout (master...fix_cleanup) http://git.io/NNSs
14:22 ingy back
14:22 Mithaldu wb
14:23 * ingy looks back 4.6 hours
14:23 Mithaldu :)
14:23 Mithaldu fwiw, i'd still also like input on the module name there
14:27 ingy Function::IntHash::Bit96
14:27 Mithaldu cheers :)
14:27 ingy :)
14:27 ingy I like it
14:27 ingy use Inline::Module too of course
14:27 Mithaldu i mainly want it to make a starfield in a perl opengl game
14:28 ingy and we'll get through the sticky points together
14:28 * Mithaldu looks that up
14:28 Mithaldu "Make "XS" modules for CPAN. The easy way!"
14:28 Mithaldu that sounds exactly like what the doctor ordered
14:28 ingy That code was written a couple months ago
14:28 ingy under a high profile TPF grant
14:29 ingy very surprised you missed that
14:29 Mithaldu fitting that one of the first uses for it would be a completely irreverent project :D
14:29 ingy http://inline.ouistreet.com/
14:30 Mithaldu i miss a lot of things, because i generally do not make an effort to read blogs outside of bpo
14:30 ingy it was cross blogged there aften
14:30 ingy often
14:30 ingy regardless…
14:30 Mithaldu well that teaches me to open my mouth
14:30 ingy let me look at your PRs
14:31 Mithaldu so it was!
14:31 ingy also I'll likely wait for mohawk to review
14:31 ingy and I owe him a couple PR reviews
14:31 ingy but we'll get it out today
14:31 Mithaldu and they're not yet marked as read, so i saw them scroll by, but was distracted by other things
14:31 ingy :D
14:31 Mithaldu well no hurry on my account, on my box they're fixed
14:32 ingy I'm very gald that you're interested now
14:32 Mithaldu though i'll of course be happy if fixes make it out :)
14:32 Mithaldu i expected i'd be!
14:32 Mithaldu doing game stuff has a lot of math bits that need to be optimized
14:32 ingy Mithaldu + ingyware usually ≡ \o/
14:32 Mithaldu haha
14:33 Mithaldu oh wow, you wrote a lot on that report
14:34 ingy Mithaldu: in #pegex, we've be working on String::Slice, an Inline::Module module that we are trying to milk speed out of
14:34 * Mithaldu grumbles at google alerts for not sending a ping that ingy mentioned him
14:34 ingy basically making a faster substr
14:34 ingy where did I mention you? :)
14:34 Mithaldu 15:34:44 ººº¹ #pegex You have joined too many channels
14:34 Mithaldu fffff
14:34 ingy ???!
14:34 Mithaldu in the wrap up post
14:35 ingy oh yeah!
14:35 ingy you worked on that project!
14:35 Mithaldu ingy: where is #pegex?
14:36 ingy it was really a massive team effort
14:36 ingy oh sorry freenode
14:36 Mithaldu aaah
14:36 ingy anything that is not totally tied to perl goes there
14:36 Mithaldu Fractal and another person are also sitting in #pegex here
14:36 ingy even though mostly perl people to start
14:36 Mithaldu makes sense
14:37 ingy like #git-commands :)
14:37 ingy I have thus far failed to gain module popularity in the other (lesser) languages ;)
14:38 ingy well Bash is superior (obviously)
14:38 Mithaldu haha
14:38 ingy I've written a lot of bash that I would challenge anyone to write cleaner in another lang
14:38 Mithaldu urgh, the weather is very english here today
14:39 ingy uuuurgh!
14:39 Mithaldu got links?
14:39 ingy where are you?
14:39 Mithaldu i like looking at clean code even if it's french to me
14:39 ingy hmm I'll ping you when I remember something
14:39 Mithaldu have a webcam: http://www.zander-info.de/rooms.html
14:39 Mithaldu sure, no hurry
14:40 ingy are you in .de today?
14:40 Mithaldu yeah
14:40 ingy I wish I was in Berlin
14:40 Mithaldu i'm a km away fro that position across the rails
14:40 ingy guess I will be soon
14:40 Mithaldu :)
14:40 ingy Ah Hannover
14:41 ingy the halfway point from berlin to lizmat
14:41 ingy :)
14:41 ingy was just there in Oct
14:41 Mithaldu here's one that's a little closer to me: http://www.langelaube2.de/ausblick
14:41 ingy looks very english/seattle
14:42 Mithaldu virtually indistinguishable from london
14:42 Mithaldu what'd you do in hannover?
14:42 ingy passed through on a train from berlin to lizmat
14:42 Mithaldu haha
14:42 Mithaldu it's worth stopping tbh
14:43 ingy when I arrived I was informed that I had just had my TPF grant accepted
14:43 Mithaldu we have a church that was bombed and the walls remained along with the bell tower
14:43 ingy so now you can interpolate when
14:43 Mithaldu so they cleaned it up, repaired the bells and hung up a bell that was a present from hiroshima to hannover
14:43 ingy :)
14:43 Mithaldu but didn't put back a roof or anything
14:43 Mithaldu just the walls
14:44 Mithaldu time is nebulous to me to be honest :)
14:44 ingy I should change the channel topic
14:44 Mithaldu hehe
14:45 Topic for #inline is now LOG: http://irclog.perlgeek.de/inline | Discussion of Inline.pm and ILSMs | https://github.com/ingydotnet/inline-pm | http://inline.ouistreet.com
14:45 ingy ok let me look at your PRs for a few mins
14:46 ingy before I need to get to the office
14:46 ingy brb
14:46 Mithaldu aight
14:47 rurban1 joined #inline
14:48 rurban1 left #inline
14:55 GitHub8 [inline-pm] ingydotnet pushed 1 new commit to master: http://git.io/NAUP
14:55 GitHub8 inline-pm/master a1b40fd Christian Walde: avoid {}s in generated Makefile code to avoid breaking dmake...
14:55 GitHub166 [inline-pm] ingydotnet closed pull request #51: avoid {}s in generated Makefile code to avoid breaking dmake (master...patch-1) http://git.io/NN2R
14:55 ingy neato trick on PR/51
14:55 Mithaldu least complicated thing i could think of :)
14:55 Mithaldu not very readable, but better than figuring out how to get the proper quoting function out of eumm
14:57 ingy for sure. I really think that an internal function like Inline::satisfy_makefile_dep should just read args from @ARGV
14:57 ingy but it's fixed for now so…
14:58 ingy like a one liner should be: perl -MInline=satisfy_makefile_dep foo bar baz
14:58 ingy no -e
14:59 ingy for that kind of stuff
14:59 Mithaldu pretty much
14:59 ingy but I was young and dumb
14:59 ingy the young part has changed
14:59 Mithaldu haha, honestly, i had no idea it was that old
14:59 Mithaldu also
14:59 ingy Inline was my very first module
14:59 Mithaldu overall i'm impressed how 3 separate distributions colluded to obscure the root cause of this test failure
15:00 ingy :[
15:00 mohawk i do indeed have answers
15:00 mohawk eumm has a method ->oneliner that does the right thing and encodes what needs encoding
15:01 mohawk Inline.pm does a number of less than optimal things re makefiles which i've fixed in PDL's Inline::Pdlpp
15:01 ingy hi mohawk
15:01 mohawk and i'll backport to inline itself if/when ingy helps me finish up the inline::module stuff ;-)
15:01 mohawk hi ingy, hope you're well
15:02 Mithaldu ingy: no need to be sad, that wasn't criticism
15:02 Mithaldu just an amazed observation
15:02 Mithaldu also, inline has a first module is quite ambitious!
15:16 GitHub113 [inline-module-pm] ingydotnet pushed 1 new commit to master: http://git.io/NAst
15:16 GitHub113 inline-module-pm/master 4364abe Ed J: Make write_module DEBUG not give code
15:16 Mithaldu ingy: is there already some kind of dzil plugin for inline module?
15:17 GitHub69 [inline-module-pm] ingydotnet closed pull request #30: Fix DEBUG function to work not just print a number (master...fixdebug) http://git.io/bkPZ
15:17 GitHub163 [inline-c-pm] ingydotnet created PR/37 (+2 new commits): http://git.io/NAs5
15:17 GitHub163 inline-c-pm/PR/37 6911bfa Christian Walde: reenable bs/exp/lib deletion on cleanup...
15:17 GitHub163 inline-c-pm/PR/37 54de4b0 Christian Walde: add sanity checks to cleanup deletion...
15:18 GitHub25 [inline-c-pm] ingydotnet deleted PR/37 at 54de4b0: http://git.io/NAsA
15:18 GitHub154 [inline-module-pm] ingydotnet pushed 1 new commit to master: http://git.io/NAG3
15:18 GitHub154 inline-module-pm/master 64df804 Ingy döt Net: CPAN Release 0.34...
15:18 GitHub184 [inline-c-pm] ingydotnet pushed 2 new commits to master: http://git.io/NAGG
15:18 GitHub184 inline-c-pm/master 6911bfa Christian Walde: reenable bs/exp/lib deletion on cleanup...
15:18 GitHub184 inline-c-pm/master 54de4b0 Christian Walde: add sanity checks to cleanup deletion...
15:19 ingy Mithaldu: yes
15:19 Mithaldu \o/
15:20 ingy Mithaldu: look at ingydotnet/acme-math-xs-pm
15:20 Mithaldu cheers and thanks :)
15:20 ingy it has a branch for every module framework
15:20 Mithaldu thoughtful!
15:21 ingy and an Alt-Acme-Math-XS-DistZilla release on cpan
15:21 mohawk ingy, thanks for merge
15:21 ingy s/DistZilla/<something less bloated>/ for @all_frameworks;
15:21 mohawk just rebasing my fixblib now
15:21 ingy mohawk: merge and /release/ !
15:22 mohawk yes, saw that
15:22 mohawk nice :-)
15:22 Mithaldu i'm only asking for dzil because it means i can use my old workflow with a minimum of work, and then go on to making a pew pew thing
15:22 ingy ok, remind me to review that later
15:22 ingy Mithaldu: I ♥ dzil
15:23 ingy I also ♥2hate dzil ☺
15:23 Mithaldu haha
15:23 ingy dzil is a very useful monstrosity
15:23 ingy as is Inline and …
15:24 ingy Mithaldu: one thing to point out
15:24 Mithaldu i find it a tragedy that this is not in unicode: http://i.somethingawful.com/forumsystem/emoticons/emot-madmax.gif
15:24 ingy to test code with Inline::Module, just run `prove -lv t/`
15:24 ingy just like a pure perl module
15:24 Mithaldu aight, i'll remember that
15:25 ingy don't think that it's C until you need to
15:25 Mithaldu though i guess tests will take a bit since right now i'm not actually sure what that function does!
15:25 ingy heh
15:25 mohawk that's currently true with inline::module
15:25 ingy I'd be glad to review the repo when you get it done (hour or so right? :)
15:25 mohawk but i think it doesn't scale well
15:26 Mithaldu ingy: longer, and no hurry
15:26 Mithaldu i'm making food and such
15:26 mohawk if you have multiple large modules, and especially tests that do other stuff with Inline, i::m needs changes
15:26 mohawk those changes are in pr/31 ;-)
15:26 ingy that's what I was doing earlier
15:27 mohawk grin
15:27 Mithaldu ingy: all i wanted to do was to implement this algorithm in perl: http://nullprogram.com/blog/2011/06/13/
15:27 mohawk i know, just giving Mithaldu an insight
15:27 Mithaldu and now it's 5 hours later! :D
15:27 mohawk ingy, i've rebased pr/31
15:27 ingy mohawk: I meant making food
15:27 mohawk grin
15:27 mohawk i've just finished eating as well
15:28 ingy Mithaldu: that's not too bad (for programming)
15:28 ingy I need to get my body to the office now.
15:28 ingy ☺/
15:29 leont ingy: trying out your git-hub tool, but getting an «Error: 'git hub repo-new' command failed: 404» :-s
15:30 Mithaldu good trip o/
15:32 travis-ci ingydotnet/inline-c-pm#139 (PR/37 - 54de4b0 : Christian Walde): The build has errored.
15:32 travis-ci Change view : https://github.com/ingydotnet/inline-c-pm/compare/6911bfa95c7c^...54de4b060abf
15:32 travis-ci Build details : http://travis-ci.org/ingydotnet/inline-c-pm/builds/51090387
15:33 Mithaldu noooooo D:
15:34 Mithaldu oh that is because travis is borken it seems
15:34 mohawk oh dear
15:35 mohawk holy cow, the code used to say:
15:35 mohawk File::Spec->catfile($autodir,'$modfname.bs'),
15:35 mohawk single-quotes??!!
15:35 mohawk how the christ did that ever work?
15:36 Mithaldu probably not
15:41 davido_ltp joined #inline
15:44 davido_ltp joined #inline
15:45 davido___ joined #inline
16:29 ingy the travis fail was from me pushing a wrong branch then deleting it immediately
16:29 Mithaldu ingy: yooooo that branch arrangement is clever and looks hilarious in tortoise
16:30 ingy ?
16:30 Mithaldu acme math xs
16:30 ingy oh :)
16:30 ingy I think all the branches are orphaned from each other
16:31 Mithaldu yes!
16:31 ingy you can keep all the branches of every git repo ever made in one git repo
16:32 Mithaldu ingy: basically this was my reaction to seeing someone else to that: https://www.youtube.com/watch?v=dHY5ZBSHzyw&amp;list=PL57hJfweW_2sv1WtKWyFiua5r0QivDTZl#t=1348
16:32 ingy I use that trick to get past the 5 repo private limit on GitHub
16:32 Mithaldu ahahaha, that's cute
16:34 ingy it's not even really a trick, one you see git history chains in that light
16:34 Mithaldu i know, i've used that a LONG time ago in some project to store downloads :)
16:39 ingy Mithaldu: https://travis-ci.org/ingydotnet/inline-c-pm/builds
16:40 ingy passing.
16:40 ingy I should config travis to ignore certain branches
16:41 Mithaldu \o/
16:41 ingy mohawk: can you review Mithaldu's inline{,-c}-pm commits
16:41 ingy then I'll release them
16:41 mohawk ingy, i will do so later today
16:41 ingy thanks
16:43 mohawk ingy, please ponder my fixblib pr ;-)
16:50 GitHub111 [inline-c-pm] wchristian comment on issue #36: @ingydotnet : This should not be closed yet unless the dep version on Inline.pm has been bumped. ;) http://git.io/Nxfb
16:51 mohawk ingy, issues currently: i don't know very much about MB so i don't know how much (if any) work would be needed for that
16:51 mohawk i suspect none since it just blindly rebuilds every time, i think
16:52 mohawk (but then inline doesn't)
16:52 mohawk and also, i don't know how much doc-age you'd want for it - probably a few words in the tutorial?
18:17 Mithaldu ingy: i thought Inline::Module would convert it into an xs module when doing build?
18:19 Mithaldu ah, the dll is made quietly on make, neat
18:39 ingy Mithaldu: the overall process of Inline::Module is still evolving
18:40 ingy we wanted to make sure it 1) worked everywhere 2) didn't dep (on user side) on anything that XS wouldn't dep on
18:40 Mithaldu right
18:40 ingy regardless of whether you agree with that approach, that's the one we took
18:40 ingy so we made abuse of inc/
18:40 Mithaldu ingy: my only concern is i don't want to have users compile a thing whenever they start it :v
18:41 ingy after install, it should be indistinguishable from XS
18:42 ingy well maybe indistinguishable is not exactly the right word
18:42 ingy all compile happens when it should
18:42 ingy over time we will modify things to bundle less in inc/
18:43 ingy at least by default
18:43 ingy there might be situations that require it
18:43 ingy we don't know yet for sure
18:44 Mithaldu aight :)
18:45 Mithaldu it seems to work fairly smooth
18:45 Mithaldu only the quietness of the build step unnerves me
18:45 ingy yeah for sure
18:45 ingy PERL_INLINE_MODULE_DEBUG=1
18:46 Mithaldu i just mean during install :)
18:46 ingy make sure you have the Inline::Module I released this morning
18:46 Mithaldu i don't need it to spew all the time
18:46 ingy ok
18:46 ingy I see
18:46 ingy yeah, we wanted it to feel more like pure perl
18:46 Mithaldu i don't mind env vars, i got the strictures one set permanently
18:46 Mithaldu yeah
18:47 Mithaldu i guess it's a matter of taste
18:47 Mithaldu i like CPAN.pm style
18:47 Mithaldu cpanm just feels wrong
18:47 ingy heh
18:47 Mithaldu like one of these cellphone apps that shows only a spinner because of course i'm a dumb user who'd be confused by it spewing debug statements when doing a long thing
18:47 ingy you want to be sprayed in the face with garbage. I understand...
18:47 Mithaldu yes
18:48 Mithaldu because then i know that's not stuck in the background
18:48 ingy well there should be options to get what you want
18:48 ingy patches welcome
18:48 Mithaldu word
23:09 GitHub194 [inline-c-pm] mohawk2 reopened issue #36: t/27inline_maker.t fails on MS Windows http://git.io/AOvugA
23:14 mohawk Mithaldu, which version does I::C depend on now?
23:15 Mithaldu whatever version's the next release of inline-pm
23:15 mohawk good shout
23:16 Mithaldu mohawk: also check #win32
23:17 mohawk Mithaldu, for what?
23:17 Mithaldu kill 0 related stuff
23:17 mohawk oh god, that
23:17 mohawk ingy, Mithaldu's work looks ok to me

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