Perl 6 - the future is here, just unevenly distributed

IRC log for #perl6-dev, 2017-10-11

| Channels | #perl6-dev index | Today | | Search | Google Search | Plain-Text | summary

All times shown according to UTC.

Time Nick Message
00:11 benchable6 joined #perl6-dev
00:26 BenGoldberg joined #perl6-dev
00:49 llfourn joined #perl6-dev
01:00 MasterDuke isn't there some reason that usually happens? fwiw, it used to just be a single method, signature of `(var, int $i, int $m, int $monkey, int $s, int $a, $context)`
01:02 MasterDuke and it's now `proto (|), (Iterable:D \var, ...), (Mu:D \var, ...), (Mu:U \var, ...)`
01:41 ugexe fwiw Net::HTTP broke sometime recently, complaining of using a `last` outside of a loop construct. as a workaround i redid the loop to not use `last`, but I also couldn't golf it
01:56 ilbot3 joined #perl6-dev
01:56 Topic for #perl6-dev is now Perl 6 language and compiler development 2.0 | Logs at http://irclog.perlgeek.de/perl6-dev/today | For toolchain/installation stuff see #perl6-toolchain | For MoarVM see #moarvm
01:59 ugexe https://raw.githubusercontent.com/ugexe/Perl6-Net--HTTP/master/xt/500-transport.t was one of the failing tests
02:00 AlexDaniel- c: HEAD https://gist.githubusercontent.com/AlexDaniel/8d9110472c53c45c327811598295a9d7/raw/927e5ec5540fff38031c3546406bb3608a193cb8/test.p6
02:00 committable6 AlexDaniel-, Successfully fetched the code from the provided URL.
02:00 committable6 AlexDaniel-, ¦HEAD(a72214c): «1..4␤===SORRY!===␤last without loop construct «exit code = 1»»
02:00 AlexDaniel- here
02:00 AlexDaniel- c: 2017.07 https://gist.githubusercontent.com/AlexDaniel/8d9110472c53c45c327811598295a9d7/raw/927e5ec5540fff38031c3546406bb3608a193cb8/test.p6
02:00 committable6 AlexDaniel-, Successfully fetched the code from the provided URL.
02:01 committable6 AlexDaniel-, https://gist.github.com/73e11623eb911ed07594a7c7afed24db
02:01 AlexDaniel- bisect: old=2017.07 https://gist.githubusercontent.com/AlexDaniel/8d9110472c53c45c327811598295a9d7/raw/927e5ec5540fff38031c3546406bb3608a193cb8/test.p6
02:01 bisectable6 AlexDaniel-, Successfully fetched the code from the provided URL.
02:01 bisectable6 AlexDaniel-, Bisecting by exit code (old=2017.07 new=a72214c). Old exit code: 0
02:01 AlexDaniel- *heavy breathing*
02:01 bisectable6 AlexDaniel-, bisect log: https://gist.github.com/5e61264647a48c145c8bf0d93178087d
02:01 bisectable6 AlexDaniel-, (2017-10-07) https://github.com/rakudo/rakudo/commit/b59804bc51b6a7f9941009fffd52957c3d408371
02:02 AlexDaniel- Geth: ver https://github.com/rakudo/rakudo/commit/b59804bc51b6a7f9941009fffd52957c3d408371
02:02 Geth AlexDaniel-, version bump brought in these changes: https://github.com/perl6/nqp/compare/2017.09-38-ga0618a6...2017.09-40-ga6a1aa0
02:02 AlexDaniel- Geth: ver https://github.com/perl6/nqp/commit/a6a1aa0b48691978237d94b0d4deb2eee98b028a
02:02 Geth AlexDaniel-, version bump brought in these changes: https://github.com/MoarVM/MoarVM/compare/2017.09.1-62-g89ca8eb...2017.09.1-553-ga4fef0b
02:02 AlexDaniel- ugexe: pretty sure it's a JIT issue looking at this bump
02:02 AlexDaniel- but could be anything, I can't bisect into moar with a bot yet
02:03 AlexDaniel- let's double-check though
02:03 AlexDaniel- c: b59804bc^,b59804bc https://gist.githubusercontent.com/AlexDaniel/8d9110472c53c45c327811598295a9d7/raw/927e5ec5540fff38031c3546406bb3608a193cb8/test.p6
02:03 committable6 AlexDaniel-, Successfully fetched the code from the provided URL.
02:03 committable6 AlexDaniel-, https://gist.github.com/d58263ac81c276d0a3c54d266220ed2e
02:03 AlexDaniel- yeah, that is it
02:03 AlexDaniel- ugexe: rakudobug plz
02:04 * AlexDaniel- &
02:09 MasterDuke .ask jnthn i've got a question about multis and find_best_dispatchee here https://irclog.perlgeek.de/perl6-dev/2017-10-10#i_15286003, with some relevant info here https://irclog.perlgeek.de/perl6-dev/2017-10-11#i_15286163
02:09 yoleaux MasterDuke: I'll pass your message to jnthn.
02:09 MasterDuke .ask timotimo i've got a question about multis and find_best_dispatchee here https://irclog.perlgeek.de/perl6-dev/2017-10-10#i_15286003, with some relevant info here https://irclog.perlgeek.de/perl6-dev/2017-10-11#i_15286163
02:09 yoleaux MasterDuke: I'll pass your message to timotimo.
02:10 MasterDuke .ask lizmat i've got a question about multis and find_best_dispatchee here https://irclog.perlgeek.de/perl6-dev/2017-10-10#i_15286003, with some relevant info here https://irclog.perlgeek.de/perl6-dev/2017-10-11#i_15286163
02:10 yoleaux MasterDuke: I'll pass your message to lizmat.
02:12 |Tux| joined #perl6-dev
03:52 evalable6 joined #perl6-dev
04:38 travis-ci joined #perl6-dev
04:38 travis-ci Rakudo build failed. Zoffix Znet 'Bump NQP'
04:38 travis-ci https://travis-ci.org/rakudo/rakudo/builds/286282249 https://github.com/rakudo/rakudo/compare/e19b3409e857...a72214c4f160
04:38 travis-ci left #perl6-dev
04:38 buggable [travis build above] ✓ All failures are due to: GitHub connectivity (1 failure).
06:04 [Tux] This is Rakudo version 2017.09-372-ga72214c4f built on MoarVM version 2017.09.1-591-gf2deedfc
06:04 [Tux] csv-ip5xs        1.214 -  1.377
06:04 [Tux] test             9.800 - 10.478
06:04 [Tux] test-t           3.215 -  3.428
06:04 [Tux] csv-parser       0.887 -  0.964
06:18 brrt joined #perl6-dev
06:23 patrickz joined #perl6-dev
07:26 brrt joined #perl6-dev
07:43 jsimonet joined #perl6-dev
09:00 robertle joined #perl6-dev
09:22 Geth ¦ nqp: 45cb3ee248 | pmurias++ | src/core/testing.nqp
09:22 Geth ¦ nqp: [jvm] Workaround a try/CATCH bug in dies-ok
09:22 Geth ¦ nqp: review: https://github.com/perl6/nqp/commit/45cb3ee248
09:22 Geth ¦ nqp: 0ccdf2e535 | pmurias++ | src/vm/jvm/runtime/org/perl6/nqp/sixmodel/reprs/P6bigintInstance.java
09:22 Geth ¦ nqp: [jvm] Throw an exception when unboxing a too big bignum to an int
09:22 Geth ¦ nqp: review: https://github.com/perl6/nqp/commit/0ccdf2e535
09:35 travis-ci joined #perl6-dev
09:35 travis-ci NQP build failed. pmurias '[jvm] Throw an exception when unboxing a too big bignum to an int'
09:35 travis-ci https://travis-ci.org/perl6/nqp/builds/286427600 https://github.com/perl6/nqp/compare/26abb152bbf3...0ccdf2e535e7
09:35 travis-ci left #perl6-dev
10:06 buggable New CPAN upload: Config-1.2.2.tar.gz by TYIL https://cpan.metacpan.org/authors/id/T/TY/TYIL/Perl6/Config-1.2.2.tar.gz
10:27 brrt joined #perl6-dev
11:20 W4RL0RD joined #perl6-dev
11:20 timotimo ===> Testing: cro:ver<0.7>
11:20 timotimo WARNING: unhandled Failure detected in DESTROY. If you meant to ignore it, you can mark it as handled by calling .Bool, .so, .not, or .defined methods. The Failure was:
11:20 timotimo No such symbol 'Cro::Tools::LinkTemplate::HTTPS'
11:20 timotimo interesting
11:59 Zoffix awhile back some user had similar issue in #perl6. It was a `require` setup (I think with a `try`) from which the unhandled Failure was popping up and I couldn't see where it'd be coming from (IIRC the module was loading properly or the failure from require was being handled)
12:00 jnthn Yeah, it's RT'd
12:00 jnthn I'm sure we're disarming any Failure return gives us back
12:00 jnthn So I think it's some Rakudo bug. Not to mention that I'm sure that symbol does exist.
12:01 timotimo it could very well be we're accidentally cloning (or something?) the failure inside the module loader?
12:01 jnthn Dunno
12:31 Geth ¦ nqp/master: 4 commits pushed by pmurias++
12:31 Geth ¦ nqp/master: 1b836b1d60 | [js] Have separate methods for different bindpos variants
12:31 Geth ¦ nqp/master: efa763bbb8 | [js] Fix cloning of a type object
12:31 Geth ¦ nqp/master: e9e338b1a0 | [js] Make nqp::getuniname spit out <noncharacter-XXXX> and <control-XXXX>
12:31 Geth ¦ nqp/master: ebd18b86e5 | Add extra multidim tests
12:31 Geth ¦ nqp/master: review: https://github.com/perl6/nqp/compare/0ccdf2e535...ebd18b86e5
12:37 Zoffix ZOFVM: Files=1277, Tests=152650, 178 wallclock secs (21.69 usr  3.68 sys + 3415.87 cusr 217.02 csys = 3658.26 CPU)
12:38 Geth ¦ rakudo/nom: fbae69e2be | (Zoffix Znet)++ | src/core/traits.pm
12:38 Geth ¦ rakudo/nom: Make `is default` on routines work earlier in setting
12:38 Geth ¦ rakudo/nom:
12:38 Geth ¦ rakudo/nom: Sometimes it's needed but the `does` op it's using is defined
12:38 Geth ¦ rakudo/nom: late in the setting.
12:38 Geth ¦ rakudo/nom:
12:38 Geth ¦ rakudo/nom: Mix in the marker role directly using .^mixin instead.
12:38 Geth ¦ rakudo/nom: review: https://github.com/rakudo/rakudo/commit/fbae69e2be
12:38 Zoffix timotimo: ^ that should make it work for you, if you still need it.
12:39 timotimo that's simple enough. thanks!
12:40 * Zoffix is failing to find any tests for this feature :(
12:41 Zoffix ah, S12-methods/default-trait.t
12:45 travis-ci joined #perl6-dev
12:45 travis-ci NQP build failed. pmurias 'Add extra multidim tests'
12:45 travis-ci https://travis-ci.org/perl6/nqp/builds/286493611 https://github.com/perl6/nqp/compare/0ccdf2e535e7...ebd18b86e58a
12:45 travis-ci left #perl6-dev
12:55 pmurias joined #perl6-dev
13:03 samcv Zoffix, how would you feel about IO::Path.parent(2) or some number for how many parent directories to go up?
13:04 samcv do you like that idea?
13:06 Geth joined #perl6-dev
13:06 Geth ¦ rakudo/nom: 714c188d4a | (Timo Paulssen)++ | src/core/native_array.pm
13:06 Geth ¦ rakudo/nom: generalize constraint for append
13:06 Geth ¦ rakudo/nom:
13:06 Geth ¦ rakudo/nom: otherwise it'll only be called when an int64 array is
13:06 Geth ¦ rakudo/nom: passed. This makes it splice 32bit, 16bit and 8bit
13:06 Geth ¦ rakudo/nom: for basically free.
13:06 Geth ¦ rakudo/nom: review: https://github.com/rakudo/rakudo/commit/714c188d4a
13:06 timotimo spec tests were clean
13:08 Zoffix samcv: is it different from IO::Path.add('../../') ?
13:09 samcv well. that's really a unixism
13:09 Zoffix Why?
13:09 samcv well it's not as clean
13:09 samcv you want the parent to 2 levels not just 1. what if you want the 5th parent
13:09 samcv or want to do it programatically?
13:09 samcv based on depth?
13:10 Zoffix IO::Path.add('../' x $depth)
13:11 samcv just because there's one way to possibly do it doesn't mean there shouldn't be another
13:11 samcv i don't like the idea of having to manipulate the path myself
13:11 samcv by multiplying strings
13:12 samcv and if we want people to use .child, it seems to follow that having .parent and being able to do a certain number of parents would make sense
13:12 samcv why even have parent at all if you can just do .add('../') ?
13:13 Zoffix samcv: don't really care about it one way or another, but feels like a YAGNI: adding a million of options just because you thought of them. Perl's App::ZofCMS has this problem with its plugins and their docs are a mile long just because they have to document all of these options that ultimately make no difference
13:13 samcv curious what other peoples thoughts on this are?
13:15 Zoffix <Zoffix>  samcv: is it different from IO::Path.add('../../')
13:15 Zoffix Yeah, it is.
13:15 Zoffix m: .parent.parent.resolve.say with "foo".IO
13:15 camelia rakudo-moar 714c18: OUTPUT: «"/home".IO␤»
13:15 Zoffix m: .add('../../').resolve.say with "foo".IO
13:15 camelia rakudo-moar 714c18: OUTPUT: «"/home/camelia/foo/../..".IO␤»
13:15 Zoffix needs to be .parent.add('../' x $depth-1) to be the same
13:16 samcv m: say $*CWD.parent.add('../')
13:16 camelia rakudo-moar 714c18: OUTPUT: «"/home/../".IO␤»
13:16 Zoffix Ah, not even then. Looks like .parent is not symlink safe, while .add is
13:17 Zoffix m: my $p = "foo/bar/ber/meows".IO; my $depth = 3; $p .= parent xx $depth; say 4p
13:17 camelia rakudo-moar 714c18: OUTPUT: «5===SORRY!5=== Error while compiling <tmp>␤Confused␤at <tmp>:1␤------> 3depth = 3; $p .= parent xx $depth; say 47⏏5p␤»
13:17 Zoffix m: my $p = "foo/bar/ber/meows".IO; my $depth = 3; $p .= parent xx $depth; say $p
13:17 camelia rakudo-moar 714c18: OUTPUT: «"foo".IO␤»
13:18 Zoffix m: my $p = "foo/bar/ber/meows".IO; my $depth = 3; $p .= add('../') xx $depth; say $p
13:18 camelia rakudo-moar 714c18: OUTPUT: «"foo/bar/ber/meows/../../../".IO␤»
13:18 Zoffix samcv: well, you got the commit bit :) I have no opinion on this
13:18 * Zoffix &
13:18 samcv ok :)
13:19 samcv was also asking if adding just an int argument was fine with you, since you had done work on the IO operations arguments as well
13:19 samcv though i think .parent a number by itself not a named argument makes most sense
13:20 samcv also this:
13:20 samcv m: $*CWD.parent.add('../').absolute.say
13:20 camelia rakudo-moar 714c18: OUTPUT: «/home/..␤»
13:21 samcv so it seems to be different in general and i didn't know for sure about the symlink thing, but i had an inkling there was some difference
13:26 stmuk_ joined #perl6-dev
13:33 [Coke] ff
13:40 * [Coke] is using rakudo for some devops stuff at work and is happy.
13:49 brrt left #perl6-dev
13:50 pmurias jnthn: is there anything besides BUILDALL on the jvm backend that might be affected by a change to BUILDALLPLAN?
13:57 timotimo lizmat: would be interested to hear startup timings with my last few moarvm commits
14:14 * Zoffix bumps
14:15 Zoffix samcv: yeah, int argument sounds fine
14:15 AlexDaniel :o even better startup times? :o :o
14:15 Zoffix m: $*CWD.parent.add('../').resolve.absolute.say
14:15 camelia rakudo-moar 714c18: OUTPUT: «/␤»
14:15 Zoffix m: 'foo'.IO.parent.parent.say
14:15 camelia rakudo-moar 714c18: OUTPUT: «"..".IO␤»
14:16 Zoffix m: 'foo'.IO.parent.parent.resolve.say
14:16 camelia rakudo-moar 714c18: OUTPUT: «"/home".IO␤»
14:16 Zoffix m: 'foo'.IO.add('../..').resolve.say
14:16 camelia rakudo-moar 714c18: OUTPUT: «"/home/camelia/foo/../..".IO␤»
14:17 dogbert2 joined #perl6-dev
14:17 Zoffix Basically .parent just chops off the path to get its value, while the .add() version adds the ../ things, which will stay if they can't be resolved. Were `foo` a symlink to some other dir, the result .parent could be said to be "wrong", as the '../' things would lead up the original dir. But I think doing `cd ..` in the shell would have the behaviour like what `.parent` does, so it makes sense to have two
14:17 Zoffix versions to do this
14:22 Zoffix ZOFVM: Files=1277, Tests=152650, 153 wallclock secs (22.21 usr  3.38 sys + 3336.72 cusr 180.39 csys = 3542.70 CPU)
14:22 Zoffix Felt like tests were flying through a lot faster (need to take care of the few tests that sit there and hog everything eventually...)
14:22 Zoffix ZOFFLOP: t/spec/S17-procasync/kill.t
14:23 Geth ¦ nqp: e9bca388a7 | (Zoffix Znet)++ | tools/build/MOAR_REVISION
14:23 Geth ¦ nqp: Bump MoarVM
14:23 Geth ¦ nqp: review: https://github.com/perl6/nqp/commit/e9bca388a7
14:23 Geth ¦ nqp: version bump brought these changes: https://github.com/MoarVM/MoarVM/compare/2017.09.1-591-gf2deedf...2017.09.1-594-gb9d3f6d
14:23 Geth ¦ rakudo/nom: a0f29e0dfa | (Zoffix Znet)++ | tools/build/NQP_REVISION
14:23 Geth ¦ rakudo/nom: Bump NQP
14:23 Geth ¦ rakudo/nom: review: https://github.com/rakudo/rakudo/commit/a0f29e0dfa
14:23 Geth ¦ rakudo/nom: version bump brought these changes: https://github.com/perl6/nqp/compare/2017.09-77-g26abb15...2017.09-84-ge9bca38
14:25 Zoffix $ time ./perl6 -e '' real0m0.122s            $ time perl6 -e '' real0m0.146s   with ./perl6 being HEAD and perl6 being 2017.09-87-g6bdb2dd
14:25 Zoffix m: say 146/122
14:25 camelia rakudo-moar 714c18: OUTPUT: «1.196721␤»
14:25 Zoffix sweet
14:26 buggable New CPAN upload: TimeUnit-0.1.0.tar.gz by ATROXAPER https://cpan.metacpan.org/authors/id/A/AT/ATROXAPER/Perl6/TimeUnit-0.1.0.tar.gz
14:31 Zoffix m: say 0.024*1277
14:31 camelia rakudo-moar 714c18: OUTPUT: «30.648␤»
14:31 Zoffix So that shaved off 30s off the spectest (were it run on 1 core)
14:31 Zoffix off *stresstest
14:35 skids joined #perl6-dev
14:38 AlexDaniel :o
14:38 timotimo what did?
14:38 AlexDaniel timo*++
14:38 timotimo who, me?
14:39 Zoffix Yeah :)
14:39 timotimo whoa
14:39 Zoffix 20% off startup
14:39 AlexDaniel Plz more? :)
14:39 timotimo it didn't make it much faster on my machine :(
14:40 timotimo still 0.10s for every empty program
14:40 Zoffix perl -e '' is 0.003 :o
14:40 timotimo no way
14:40 Zoffix On my box, yeah
14:40 timotimo okay it goes down to 0.09 more than half the time now
14:41 Zoffix on same box perl6 -e '' is 0.177s  this is 2017.09-223-gd565ded tho
14:41 AlexDaniel It is, yeah
14:41 Zoffix m: say .177/.003
14:41 camelia rakudo-moar a0f29e: OUTPUT: «59␤»
14:42 timotimo oh, sorry, i thought perl6 -e '' was 0.003 on your box
14:42 travis-ci joined #perl6-dev
14:42 travis-ci NQP build failed. Zoffix Znet 'Bump MoarVM'
14:42 travis-ci https://travis-ci.org/perl6/nqp/builds/286545373 https://github.com/perl6/nqp/compare/ebd18b86e58a...e9bca388a765
14:42 travis-ci left #perl6-dev
14:43 timotimo but it was perl5
14:43 AlexDaniel At this point we should be comparing it with python maybe
14:43 Zoffix time (for i in {1..85}; do perl -e ''; done) runs the same time as perl6 -e ''
14:43 Zoffix :)
14:44 Zoffix python takes .022s
14:44 timotimo how much time is one-core-stresstest on that same box, Zoffix?
14:45 timotimo 1277 seconds, is that what?
14:45 Zoffix `time (for i in {1..9}; do python -c ''; done)` takes same time a `perl6 e ''`... So we're within an order of magnitude. Pretty good!
14:45 AlexDaniel- yeah it's getting closer
14:46 AlexDaniel- a few years ago it was unbearable
14:46 Zoffix timotimo: dunno, I never ran one-core stresstest. 1277 is the number of files in the stresstest, multiplied by the number of miliseconds the HEAD shaves off compared to older perl6
14:46 Zoffix off start time
14:48 timotimo oh, ok
14:48 Zoffix That was for Python 2.7.3 btw
14:52 travis-ci joined #perl6-dev
14:52 travis-ci Rakudo build passed. Zoffix Znet 'Make `is default` on routines work earlier in setting
14:52 travis-ci https://travis-ci.org/rakudo/rakudo/builds/286496865 https://github.com/rakudo/rakudo/compare/a72214c4f160...fbae69e2be32
14:52 travis-ci left #perl6-dev
14:52 AlexDaniel- hmmm. Here it went down from 0.161s to 0.156s
14:53 AlexDaniel- just 3% or so but it adds up
15:00 DrForr fg
15:01 Zoffix bash: fg: current: no such job
15:19 cognominal joined #perl6-dev
15:39 AlexDaniel- hmm, it's 0.140s now… nevermind
15:39 AlexDaniel- I think I had some background crap running
15:57 timotimo hum. my static js doesn't reach the browser for some reason
15:57 timotimo oh, haha
15:57 timotimo probably "cro trace"'s fault for trying to hexdump 1.2 megs of javasrcipt
16:06 buggable New CPAN upload: TimeUnit-0.1.1.tar.gz by ATROXAPER https://cpan.metacpan.org/authors/id/A/AT/ATROXAPER/Perl6/TimeUnit-0.1.1.tar.gz
16:09 timotimo huh. i start{} some stuff from the mainline of the main file but they only get run after i hit ctrl-c?
16:11 ggoebel joined #perl6-dev
16:39 robertle joined #perl6-dev
16:51 AlexDaniel- timotimo: are you sure it's not output buffering?
16:55 travis-ci joined #perl6-dev
16:55 travis-ci Rakudo build passed. Timo Paulssen 'generalize constraint for append
16:55 travis-ci https://travis-ci.org/rakudo/rakudo/builds/286508647 https://github.com/rakudo/rakudo/compare/fbae69e2be32...714c188d4aa0
16:55 travis-ci left #perl6-dev
16:57 AlexDaniel- Zoffix: so what if I move al 「testneeded」 tickets to roast issues and we make a roast squashathon next month?
16:57 AlexDaniel- does it make sense?
16:59 timotimo https://imgur.com/CziybHs
16:59 timotimo AlexDaniel-: it was in fact output buffering :)
17:01 Zoffix AlexDaniel-: I'm a bad person to ask this, because I don't like most of the PRs for tests :}
17:01 Zoffix buggable: zen TESTING IS AN ART!
17:01 buggable Zoffix, "When an ordinary man attains knowledge, he is a sage; when a sage attains understanding, he is an ordinary man."
17:02 Zoffix ZofBot: what about an ordinary woman?
17:02 ZofBot Zoffix, d, and who wants to do them
17:03 Zoffix ouch. Bad robot!
17:03 AlexDaniel- hah
17:03 AlexDaniel- so I couldn't sleep last night and was thinking about it
17:03 AlexDaniel- so let's say we start moving RT testneeded tickets to roast github issues
17:04 AlexDaniel- because, you know, squashathon and hacktoberfest and stuff, it's just so much easier this way
17:05 AlexDaniel- problem is, some of these tickets should be resolved by tests in rakudo/t
17:05 AlexDaniel- … so… at what point do we stop banging our heads against the wall and open up rakudo/issues on github?
17:05 Zoffix another problem is some of them ain't even fixed.
17:06 Zoffix Like the Duration.new(Inf) thing. The claimed fix just binds a Failure to a Rat attribute. Seems new behaviour is even worse.
17:06 AlexDaniel- well, great, then I'll move it back to RT :'(
17:07 AlexDaniel- I mean, look, I see no problem having to refer to tickets as RT #666 and GH #666
17:07 AlexDaniel- so old tickets can stay there, I'm fine with that
17:07 Zoffix and a third problem is you often need to read back like miles of text just to understand what exactly was wrong and how to properly cover it
17:07 AlexDaniel- but it'd be so much easier…
17:08 Zoffix Feels like we need some TESTNEEDED task force and people who like writing tests close those tickets in higher priority, rather than leaving them off as LHF for newcomers to hack on
17:08 AlexDaniel- hmmm
17:08 Zoffix I mean as a general thing, not as an argument against putting them for squashton
17:09 Zoffix buggable: tag testneeded
17:09 buggable Zoffix, There are 55 tickets tagged with TESTNEEDED; See http://fail.rakudo.party/t/TESTNEEDED for details
17:09 * Zoffix puts money where the mouth is and goes to close a couple
17:10 AlexDaniel- my problem right now is that RT makes it a nightmare for me
17:10 AlexDaniel- I have to jump through all of these hoops just because it is RT
17:11 Zoffix lizmat doesn't like RT either... Seems to me it's 2 against RT and no one actively objecting :)
17:12 Zoffix objecting migration I mean
17:12 AlexDaniel- and my proposed solution is really that simple – just *open* rakudo/issues and see how it goes. We already use “RT” prefix for tickets consistently, I see no problem whatsoever
17:12 AlexDaniel- so I'm not proposing the massive migration just yet
17:13 pmurias Zoffix: what do you want to migrate RT to?
17:14 Zoffix pmurias: I have no opinion on this matter. I'm a silent-ish observer :)
17:14 Zoffix AlexDaniel-: you got the keys to enable it tho, don't you?
17:14 AlexDaniel- the keys were not given for that purpose
17:14 Zoffix :)
17:14 * Zoffix &
17:18 AlexDaniel- lizmat: what do you think about it? ↑
17:19 AlexDaniel- do we even have anybody who is against using some other issue tracker in parallel to RT?
17:20 AlexDaniel- and by “anybody” I'm thinking someone who is relatively active. I touch RT every day and I'm so sick of it. If someone never has to interact with it and they're against the change, I couldn't care less.
17:30 Geth ¦ roast: 6e6045b783 | (Zoffix Znet)++ | S12-meta/exporthow.t
17:30 Geth ¦ roast: Test SUPERSEDE::class can be exported
17:30 Geth ¦ roast:
17:30 Geth ¦ roast: RT#132236: https://rt.perl.org/Ticket/Display.html?id=132236
17:30 Geth ¦ roast: Rakudo fix: https://github.com/rakudo/rakudo/commit/dd943eded83edb3753
17:30 Geth ¦ roast: review: https://github.com/perl6/roast/commit/6e6045b783
17:30 synopsebot RT#132236 [open]: https://rt.perl.org/Ticket/Display.html?id=132236 [REGRESSION] Meta object construction
17:32 Zoffix Hm. RT#132083 is marked as resolved but R6 still shows it
17:32 synopsebot RT#132083 [resolved]: https://rt.perl.org/Ticket/Display.html?id=132083 [REGRESSION] Broken math (-2147483648 != -2147483648)
17:33 * Zoffix clicks the "Delete" button
17:33 [Coke] I'm against a full scale migration. I'm fine with opening a second one and letting RT eventually run out of live tickets.
17:34 [Coke] github issues ain't great, but it's not horrible.
17:34 gfldex I strongly dislike RT but I do like to get e-mails when tickets I created are closed. So I'm leaning towards slow transition.
17:35 Zoffix This is one of the examples I meant by how it's hard to know what to test for: https://rt.perl.org/Ticket/Display.html?id=131924#ticket-history
17:35 gfldex however, I'm all for making work smoother for those who do close those tickes
17:36 Zoffix Especially since precomp is involved and to verify my test covers the bug I have to do gymnastics with committable6 or build the buggy commit.
17:47 Geth ¦ roast: 807ed8137f | (Zoffix Znet)++ | S32-io/dir.t
17:47 Geth ¦ roast: Test concurrentyl using dir() works
17:47 Geth ¦ roast:
17:47 Geth ¦ roast: RT#129845: https://rt.perl.org/Ticket/Display.html?id=129845
17:47 Geth ¦ roast: Rakudo fix: https://github.com/rakudo/rakudo/commit/231cb3f
17:47 Geth ¦ roast: review: https://github.com/perl6/roast/commit/807ed8137f
17:47 synopsebot RT#129845 [open]: https://rt.perl.org/Ticket/Display.html?id=129845 [TESTNEEDED] [CONC] `.dir` returns corrupted `IO::Path`s under concurrent load
17:48 Zoffix umm
17:48 Zoffix it's only ~.3s but I guess since it's just covering a non-dir-related bug...
17:48 Zoffix .in 4hr move this to stresstest https://github.com/perl6/roast/commit/807ed8137f
17:48 yoleaux Zoffix: I'll remind you at 21:48Z
17:53 timotimo Zoffix: i've tried the "say 'a' a million times" thing with "say 1" instead - in order to not do the braid switching - and it seems to make barely any difference at all in performance
17:54 Zoffix timotimo: ah, too bad then.
17:54 timotimo ;(
17:55 timotimo 5.37% time spent in or under find_symbol
17:56 timotimo MATCH is responsible for 9.28% time but it also has 560k entries compared to 60k find_symbol
17:56 W4RL0RD joined #perl6-dev
17:56 timotimo visit_children from the optimizer also spends about 8.9% of the whole program time
17:57 timotimo and 4.84% is under !alt
17:57 timotimo !cursor_pass (which i believe is also responsible for calling action methods) has an inclusive time of 28%
17:58 timotimo but funnily enough the first entry from Actions.nqp is at 8.17% inclusive (1.3% exclusive) time - it's term:sym<name>
18:02 Zoffix With `echo -n` echoing 10000x of these lines and measuring with `time perl6 -c` I get: `1.say;` = 6.9s; `1;` = 3.5s; `;` = 0m0.730s
18:02 timotimo i wonder if we have a "shortcut" for ; separated by whitespace
18:03 Zoffix Ah prolly
18:06 AlexDaniel- Zoffix: awwww I remember that ticket
18:07 AlexDaniel- maybe I can find the one liner that reproduces the issue… let's see…
18:07 timotimo only say(); over and over gets me to 4.37s (with say 1; it was 6.5s)
18:08 Zoffix AlexDaniel-: what issue?
18:08 AlexDaniel- Zoffix: RT #131924. Ah also, I don't think you have to care about precomp when using committable
18:08 synopsebot RT#131924 [new]: https://rt.perl.org/Ticket/Display.html?id=131924 [REGRESSION] \n\r and string concatenation improvements
18:08 Zoffix Ah
18:08 AlexDaniel- Zoffix: I never did and it always works, I still don't know why
18:08 AlexDaniel- I guess precomp files are different depending on what version you use? No idea actually…
18:09 Zoffix AlexDaniel-: I meant the bug manifests itself only in precomp
18:09 timotimo i get 4.06s with code like "sub boop { say 'a' };" and then only "boop;" lines
18:09 AlexDaniel- hmmmmmm oh.
18:09 timotimo also: it's somewhat hilarious to run a file that consists only of "boop;" lines without the declaration
18:10 timotimo https://imgur.com/ebkdPZQ
18:11 AlexDaniel- pwahahahah :D
18:11 masak next up: compiler error art
18:11 Zoffix timotimo: what did you use to get your timings?
18:11 timotimo "time"
18:11 timotimo the time that fish gives me
18:12 Zoffix I meant for "MATCH is responsible for 9.28% time"
18:12 timotimo oh, that was --profile-compile
18:12 Zoffix Ah. I get 0.26%
18:12 Zoffix oh nm, there's more than one entry
18:12 Zoffix (I'm painfully navigating the callgraph :P)
18:13 timotimo yeah, you should use the routines tab instead and sort by "exclusive" time :)
18:14 Zoffix ok :)
18:14 AlexDaniel- c: 5363a0742d9e^,5363a0742d9e my $a = "a" ~ "\n\r"; my $b = "\n\r"; say ($a ~ $b).chars
18:14 committable6 AlexDaniel-, ¦5363a0742d9e^: «Cannot find this revision (did you mean “e73ff23”?)» ¦5363a07: «Cannot find this revision (did you mean “e192924”?)»
18:14 AlexDaniel- :|
18:14 AlexDaniel- that's a moar sha
18:15 AlexDaniel- c: 2017.08~100,HEAD my $a = "a" ~ "\n\r"; my $b = "\n\r"; say ($a ~ $b).chars
18:15 committable6 AlexDaniel-, ¦2017.08~100,HEAD(a0f29e0): «4»
18:15 AlexDaniel- c: 2017.08~50,HEAD my $a = "a" ~ "\n\r"; my $b = "\n\r"; say ($a ~ $b).chars
18:15 committable6 AlexDaniel-, ¦2017.08~50,HEAD(a0f29e0): «4»
18:15 AlexDaniel- c: 2017.08~200,HEAD my $a = "a" ~ "\n\r"; my $b = "\n\r"; say ($a ~ $b).chars
18:15 committable6 AlexDaniel-, ¦2017.08~200,HEAD(a0f29e0): «4»
18:15 * AlexDaniel- cries
18:15 Zoffix c: releases my $a = "a" ~ "\n\r"; my $b = "\n\r"; say ($a ~ $b).chars
18:15 committable6 Zoffix, ¦releases (23 commits): «4»
18:15 AlexDaniel- c: b81597bd,HEAD my $a = "a" ~ "\n\r"; my $b = "\n\r"; say ($a ~ $b).chars
18:15 committable6 AlexDaniel-, ¦b81597b,HEAD(a0f29e0): «4»
18:15 jnthn huh, surely that should be 3...
18:16 timotimo is that what you mean when you say \r gets turned into \n?
18:16 timotimo jnthn: did you see it's \n\r rather than \r\n?
18:16 jnthn Yes, but
18:16 Zoffix But when you concat you get \r\n in the middle
18:16 timotimo yeah
18:16 AlexDaniel- a \n \r\n \r
18:17 jnthn Oh!
18:17 jnthn I didn't see the a at the start
18:17 Zoffix ?
18:17 Zoffix Ah :)
18:17 jnthn So yeah, 4 is right
18:17 AlexDaniel- ok that's not the right snippet I think
18:17 Zoffix AlexDaniel-: doesn't the fix say "this only was problematic with precompilation"?
18:18 * jnthn blames post-dinner carb crash :)
18:18 * timotimo is pre-dinner-grocery-shopping even :|
18:18 AlexDaniel- Zoffix: of course it does, but I'm blind
18:18 Zoffix What does empty string do in grammar? https://github.com/rakudo/rakudo/blob/nom/src/Perl6/Grammar.nqp#L1237
18:19 Zoffix What does <!!{}> do in gammar?
18:19 tyil confuse people
18:19 jnthn timotimo: Ah, got delivery :)
18:19 jnthn Lazy Wednesday :)
18:19 timotimo not a bad idea in itself
18:20 jnthn Zoffix: Note that it's a rule
18:20 Zoffix Ah. OK. What about <!!{}?
18:20 jnthn Zoffix: Everything before that point is code decls and adverbs, so don't trigger sigspace. The '' is to force it
18:20 timotimo "4 years ago" :D
18:21 Zoffix `<!!{ $*LANG := $*LEAF := $/.clone_braid_from(self); 1 }>`
18:21 timotimo indeed, the commit that introduced that line was all about turning tokens into rules
18:21 timotimo https://github.com/rakudo/rakudo/commit/37acfa1d76ab7bf00716b8597aa81ea783c61499
18:21 jnthn Oh, wich something in it
18:21 jnthn So a negative lookahed is not an LTM terminator
18:22 jnthn While a positive lookahead or other code block is
18:22 timotimo what a nasty trick :)
18:22 jnthn So that's a sneaky way to run some code without it having any implication for LTM
18:22 Zoffix Oh lol :)
18:22 jnthn Yeah, :my $ = ...the code...; is the other trick :P
18:22 AlexDaniel- what about just {} ?
18:23 timotimo i wonder how completely that's going to break a future refactoring that passes LTM results down into recursive rules etc
18:23 timotimo well, { } immediately terminates LTM
18:23 jnthn AlexDaniel-: {} is an LTM terminator
18:29 AlexDaniel- e: run ‘wget’, ‘https://gist.githubusercontent.com/AlexDaniel/b18897504955df0e64343904b0dcba59/raw/df8ca79ceed5b6d745d383b970139e74b477ac6c/Header.pm6’, ‘-O’, ‘sandbox/Header.pm6’
18:29 evalable6 AlexDaniel-, rakudo-moar a0f29e0df: OUTPUT: «--2017-10-11 20:29:15--  https://gist.githubusercontent.com/AlexDaniel/b18897504955df0e643…»
18:29 evalable6 AlexDaniel-, Full output: https://gist.github.com/9a7c07b87ff102c5f65fb528fd1460d4
18:29 AlexDaniel- e: run ‘wget’, ‘https://gist.githubusercontent.com/AlexDaniel/b18897504955df0e64343904b0dcba59/raw/df8ca79ceed5b6d745d383b970139e74b477ac6c/Simple.pm6’, ‘-O’, ‘sandbox/Simple.pm6’
18:29 evalable6 AlexDaniel-, rakudo-moar a0f29e0df: OUTPUT: «--2017-10-11 20:29:40--  https://gist.githubusercontent.com/AlexDaniel/b18897504955df0e643…»
18:29 evalable6 AlexDaniel-, Full output: https://gist.github.com/3aa1874910e113456c986ca97fa738d8
18:30 AlexDaniel- c: HEAD https://gist.githubusercontent.com/AlexDaniel/b18897504955df0e64343904b0dcba59/raw/9b2404da03d0f822cf7ba4996a56b3b1d1a826de/basic.p6
18:30 committable6 AlexDaniel-, Successfully fetched the code from the provided URL.
18:30 committable6 AlexDaniel-, ¦HEAD(a0f29e0): «Default constructor for 'Simple' only takes named arguments␤  in block <unit> at /tmp/fdlvKoHhpc line 7␤ «exit code = 1»»
18:30 AlexDaniel- :|
18:31 AlexDaniel- oh the test file is wrong
18:31 AlexDaniel- actually, I really don't know what this whole thing is about anymore… nevermind I guess
18:32 AlexDaniel- c: HEAD use lib ‘sandbox’; use Simple; say buggy-str
18:32 committable6 AlexDaniel-, ¦HEAD(a0f29e0): «: ␤␤»
18:33 AlexDaniel- c: HEAD use lib ‘sandbox’; use Simple; say buggy-str eq “: \n\r\n\r”
18:33 committable6 AlexDaniel-, ¦HEAD(a0f29e0): «===SORRY!=== Error while compiling /tmp/59bVh9RBUf␤Undeclared routine:␤    eq used at line 1␤ «exit code = 1»»
18:33 AlexDaniel- c: HEAD use lib ‘sandbox’; use Simple; say buggy-str() eq “: \n\r\n\r”
18:33 committable6 AlexDaniel-, ¦HEAD(a0f29e0): «True»
18:33 AlexDaniel- c: b81597bd867a use lib ‘sandbox’; use Simple; say buggy-str() eq “: \n\r\n\r”
18:33 committable6 AlexDaniel-, ¦b81597b: «False»
18:33 AlexDaniel- \o/ \o/ \o/ \o/
18:33 AlexDaniel- Zoffix: yaayayay!!
18:34 AlexDaniel- c: b81597bd867a use lib ‘sandbox’; use Simple; my $x = buggy-str(); say $x; say $x eq “: \n\r\n\r”
18:34 committable6 AlexDaniel-, ¦b81597b: «: ␤␤␤False»
18:34 AlexDaniel- ok that doesn't help
18:34 AlexDaniel- but yeah, that *is* the test case
18:35 evalable6 joined #perl6-dev
18:35 Zoffix AlexDaniel-: what's "Simple"?
18:35 AlexDaniel- Zoffix: unit class Simple; sub buggy-str is export { “: {‘’}\n\r” ~ “\n\r” }
18:36 Zoffix .oO( fetch all the files from multi-file gist )
18:36 AlexDaniel- you don't need one of the files!
18:37 Zoffix But you got Simple.pm6 on the bot box, dinnitu?
18:37 Ven joined #perl6-dev
18:37 AlexDaniel- Zoffix: I used 「run ‘wget’, …」 above to get it, yeah
18:37 Zoffix AlexDaniel-: is roast/t/spec/packages/Test/Util.pm somewhere on the box?
18:37 AlexDaniel- Zoffix: I've left a comment here: https://rt.perl.org/Ticket/Display.html?id=131924#txn-1498385
18:39 AlexDaniel- e: run ‘git’, ‘clone’, ‘https://github.com/perl6/roast.git’, ‘sandbox/roast’
18:39 Zoffix e: run «wget https://raw.githubusercontent.com/perl6/roast/master/packages/Test/Util.pm -O sandbox/Test/Util.pm»
18:39 evalable6 AlexDaniel-, rakudo-moar a0f29e0df: OUTPUT: «Cloning into 'sandbox/roast'...»
18:39 evalable6 Zoffix, rakudo-moar a0f29e0df: OUTPUT: «(exit code 1) sandbox/Test/Util.pm: No such file or directory␤The spawned command 'wget' exited unsuccessfully (exit code: 1)␤  in block <unit> at /tmp/Rl5DqFmior line 1␤»
18:39 AlexDaniel- will probably not make it in 10 seconds though
18:39 Zoffix e: use lib <sandbox/roast/packages/>; use Test::Util; say make-temp-dir
18:39 evalable6 Zoffix, rakudo-moar a0f29e0df: OUTPUT: «"/tmp/perl6_roast_4uf4VGBnNR_line1_0_5513874120385511507747188".IO»
18:39 Zoffix \o/
18:39 Zoffix neat
18:40 AlexDaniel- but you have ssh access to it, you can do whatever you want
18:42 AlexDaniel- Zoffix: fwiw I think your last wget failed because sandbox/Test directory does not exist
18:42 Zoffix mhm
18:42 AlexDaniel- e: mkdir ‘sandbox/Test’; run «wget https://raw.githubusercontent.com/perl6/roast/master/packages/Test/Util.pm -O sandbox/Test/Util.pm»
18:42 evalable6 AlexDaniel-, rakudo-moar a0f29e0df: OUTPUT: «--2017-10-11 20:42:55--  https://raw.githubusercontent.com/perl6/roast/master/packages/Tes…»
18:42 evalable6 AlexDaniel-, Full output: https://gist.github.com/bb2582feeb6b93d9bec37c94200968fc
18:43 AlexDaniel- yep
18:56 Geth ¦ roast: 28481cea18 | (Zoffix Znet)++ | S10-packages/precompilation.t
18:56 Geth ¦ roast: No funny business in precomped string strands
18:56 Geth ¦ roast:
18:56 Geth ¦ roast: RT#131924: https://rt.perl.org/Ticket/Display.html?id=131924
18:56 Geth ¦ roast: MoarVM fix: https://github.com/MoarVM/MoarVM/commit/5363a0742
18:56 Geth ¦ roast: review: https://github.com/perl6/roast/commit/28481cea18
18:56 synopsebot RT#131924 [new]: https://rt.perl.org/Ticket/Display.html?id=131924 [REGRESSION] \n\r and string concatenation improvements
18:58 AlexDaniel- Zoffix++
18:59 AlexDaniel- Zoffix: any other TESTNEEDED tickets you have problems with?
19:02 Zoffix not yet
19:07 AlexDaniel- ugexe++
19:08 Zoffix ?
19:08 Zoffix What happened?
19:08 * Zoffix is news-deprived
19:08 Zoffix Just learned that DDG has !p6mod bang
19:08 AlexDaniel- releasable6: status
19:08 releasable6 AlexDaniel-, Next release in 9 days and ≈23 hours. 1 blocker. Changelog for this release was not started yet
19:08 releasable6 AlexDaniel-, Details: https://gist.github.com/4ddde42705e2c6114d9bc62d5b3f7a37
19:09 AlexDaniel- Zoffix: he submitted that ticket ↑
19:10 Zoffix ah
19:11 Zoffix So Failure.self blows up the Failures and I thought it was awesome at the time, but I find myself needing to comment why the hell I'm adding .self into places :)
19:12 Zoffix Well, I guess it's still better than not having that option, but the method doesn't make the code self-describing
19:13 Zoffix ZofBot: .if-this-is-a-Failure-then-blow-it-up-otherwise-gimme-self
19:13 ZofBot Zoffix, anyway, this is not something we can close atm
19:14 AlexDaniel- “method self”
19:14 AlexDaniel- “Defined as:”
19:14 AlexDaniel- “method exception(Failure:D: --> Failure:D)”
19:14 AlexDaniel- huh?
19:14 AlexDaniel- https://docs.perl6.org/type/Failure#method_self
19:15 Zoffix hm... t/spec/S32-io/dir.t flopped.. The new test that tests dir() does not produce crap results under load
19:15 evalable6 joined #perl6-dev
19:15 Zoffix AlexDaniel-: fixed
19:15 AlexDaniel- was just about to push… :)
19:15 Zoffix too slow
19:16 Geth ¦ rakudo/nom: 25c87d0d1c | (Zoffix Znet)++ | src/core/Duration.pm
19:16 Geth ¦ rakudo/nom: Blow up Failures in Duration.new
19:16 Geth ¦ rakudo/nom:
19:16 Geth ¦ rakudo/nom: Otherwise we end up binding it to Rat attribute and it explodes
19:16 Geth ¦ rakudo/nom: only later, producing really confusing errors
19:16 Geth ¦ rakudo/nom: review: https://github.com/rakudo/rakudo/commit/25c87d0d1c
19:18 travis-ci joined #perl6-dev
19:18 travis-ci Rakudo build passed. Zoffix Znet 'Bump NQP'
19:18 travis-ci https://travis-ci.org/rakudo/rakudo/builds/286545369 https://github.com/rakudo/rakudo/compare/714c188d4aa0...a0f29e0dfad7
19:18 travis-ci left #perl6-dev
19:19 Zoffix hm hm hm
19:21 * Zoffix ponders Duration.new with zero-denominator Rats
19:22 Zoffix I guess it's fine to leave it in. We'll just need to improve the div-by-zero error to be more useful so when a zero-denomator Rat explodes somewhere, the error doesn't talk about "using `div`"
19:23 Zoffix m: say [ $_ ≤ $_ or $_ ≥ $_ ] with Duration.new(0/0)
19:23 camelia rakudo-moar a0f29e: OUTPUT: «[False]␤»
19:23 Zoffix :)
19:23 Zoffix m: say Duration.new(0/0)
19:23 camelia rakudo-moar a0f29e: OUTPUT: «Attempt to divide by zero using div␤  in block <unit> at <tmp> line 1␤␤»
19:23 Zoffix Yeah, more like attempt to give an unhelpful error
19:24 Zoffix m: my $d = Duration.new(0/0); ␤␤␤␤␤ say "Computing";␤␤my $d2 = $d + 42; say "Done computing";␤␤␤␤␤ say $d2
19:24 camelia rakudo-moar a0f29e: OUTPUT: «Computing␤Attempt to divide by zero using div␤  in block <unit> at <tmp> line 13␤␤Done computing␤»
19:25 Zoffix m: for $*OUT, $*ERR -> { .open: .path, :w, :0buffer };  my $d = Duration.new(0/0); ␤␤␤␤␤ say "Computing";␤␤my $d2 = $d + 42; say "Done computing";␤␤␤␤␤ say $d2
19:25 camelia rakudo-moar a0f29e: OUTPUT: «Too many positionals passed; expected 0 arguments but got 1␤  in block <unit> at <tmp> line 1␤␤»
19:26 Zoffix m: for $*OUT, $*ERR { .open: .path, :w, :0buffer };  my $d = Duration.new(0/0); ␤␤␤␤␤ say "Computing";␤␤my $d2 = $d + 42; say "Done computing";␤␤␤␤␤ say $d2
19:26 camelia rakudo-moar a0f29e: OUTPUT: «Too many positionals passed; expected 1 argument but got 2␤  in block <unit> at <tmp> line 1␤␤»
19:26 Zoffix Fine, be that way!
19:26 Zoffix m: my $d = Duration.new(0/0); ␤␤␤␤␤ note "Computing";␤␤my $d2 = $d + 42; note "Done computing";␤␤␤␤␤ note $d2
19:26 camelia rakudo-moar a0f29e: OUTPUT: «Computing␤Done computing␤Attempt to divide by zero using div␤  in block <unit> at <tmp> line 13␤␤»
19:27 Zoffix oh right, .open doesn't take a path, duh
19:29 pmurias If I want to add a test to roast for a yet unfixed bug I should fudge it?
19:29 Zoffix Yes
19:34 Geth ¦ roast: a821b71c7b | (Zoffix Znet)++ | S32-temporal/DateTime-Instant-Duration.t
19:34 Geth ¦ roast: Test Duration.new with iffy values
19:34 Geth ¦ roast:
19:34 Geth ¦ roast: RT#127341: https://rt.perl.org/Ticket/Display.html?id=127341
19:34 Geth ¦ roast:
19:34 Geth ¦ roast: - Nums representable as zero-denominator Rat work
19:34 Geth ¦ roast: - Wrong type of arg throws
19:34 Geth ¦ roast: review: https://github.com/perl6/roast/commit/a821b71c7b
19:34 synopsebot RT#127341 [open]: https://rt.perl.org/Ticket/Display.html?id=127341 [LTA] error message mentions $!tai ( Duration.new(∞) )
19:37 Geth ¦ rakudo/nom: 5d3ebc09ba | (Zoffix Znet)++ | t/05-messages/01-errors.t
19:37 Geth ¦ rakudo/nom: Test Duration.new with bad args does not mention guts
19:37 Geth ¦ rakudo/nom:
19:37 Geth ¦ rakudo/nom: RT#127341: https://rt.perl.org/Ticket/Display.html?id=127341
19:37 Geth ¦ rakudo/nom: review: https://github.com/rakudo/rakudo/commit/5d3ebc09ba
19:44 Geth ¦ nqp: usev6++ created pull request #373: [jvm] Don't wait for child when program ends
19:44 Geth ¦ nqp: review: https://github.com/perl6/nqp/pull/373
20:19 AlexDaniel- .tell jnthn Can we open rakudo/issues on github? See https://irclog.perlgeek.de/perl6-dev/2017-10-11#i_15289848 and the start of the discussion on https://irclog.perlgeek.de/perl6-dev/2017-10-11#i_15289680
20:19 yoleaux AlexDaniel-: I'll pass your message to jnthn.
20:21 Geth ¦ nqp/jvm_killprocasync: c5f4c14055 | usev6++ | 2 files
20:21 Geth ¦ nqp/jvm_killprocasync: [jvm] Support signals 1, 9, 15 for killprocasync
20:21 Geth ¦ nqp/jvm_killprocasync:
20:21 Geth ¦ nqp/jvm_killprocasync: Since SIGHUP (1) is the default signal used in Rakudo when calling
20:21 Geth ¦ nqp/jvm_killprocasync: nqp::killprocasync we have to cheat and pretend we got a SIGTERM.
20:21 Geth ¦ nqp/jvm_killprocasync:
20:21 Geth ¦ nqp/jvm_killprocasync: The only other signal supported on the JVM is SIGKILL (used under
20:21 Geth ¦ nqp/jvm_killprocasync: the hood by destroyForcibly). For all remaining signals we throw an
20:21 Geth ¦ nqp/jvm_killprocasync: Exception.
20:21 Geth ¦ nqp/jvm_killprocasync: review: https://github.com/perl6/nqp/commit/c5f4c14055
20:21 stmuk joined #perl6-dev
20:35 bartolin AlexDaniel-: one thought that crossed my mind was that it's somewhat difficult for non-insiders to choose the right place to open an issue: roast, rakudo, nqp, doc, moarvm ... IMHO it would be nice to have one place where different types of new issues are accepted (like RT was before).
20:36 bartolin AlexDaniel-: maybe something like https://github.com/google/github-issue-mover can be used to move issues to the appropriate repo (if needed). (I haven't tested that issue mover, but maybe we can find a good tool for something like that.)
20:38 bartolin AlexDaniel-: all in all I'd be fine with using github issues for new bug reports
20:38 AlexDaniel- bartolin: is it really that big of a problem for non-insiders? There are only two choices: rakudo and doc
20:39 AlexDaniel- moar and nqp problems are still rakudo issues and we want to have a ticket for that anyway, I think
20:43 bartolin well, it wasn't meant as an objection. I just thought about it when you talked about making rt tickets to roast issues.
20:43 AlexDaniel- right
20:46 pmurias bartolin: the BUILDPLAN changes are causing a test failure on the JVM backend, but I can't figure out why
20:49 bartolin pmurias: the failure in t/nqp/060-bigint.t ?
20:49 * bartolin builds a fresh nqp
20:55 pmurias bartolin: yep
20:57 stmuk_ joined #perl6-dev
21:17 cognominal joined #perl6-dev
21:18 Geth ¦ roast: 906d96ede1 | (Zoffix Znet)++ | S32-str/Collation.t
21:18 Geth ¦ roast: Add "v" to "v6.d" to avoid potential grep missage
21:18 Geth ¦ roast: review: https://github.com/perl6/roast/commit/906d96ede1
21:18 Zoffix # TODO add test with "\c[woman facepalming]", "\c[man facepalming]"
21:18 Zoffix Test for what? Any idea?
21:18 Zoffix bottom of S32-str/Collation.t
21:21 Geth ¦ roast: 5091bdd995 | (Zoffix Znet)++ | S32-str/Collation.t
21:21 Geth ¦ roast: Test collation with 'A' vs 'a'
21:21 Geth ¦ roast:
21:21 Geth ¦ roast: RT#132216: https://rt.perl.org/Ticket/Display.html?id=132216
21:21 Geth ¦ roast: MoarVM fix: https://github.com/MoarVM/MoarVM/commit/e9d331d4
21:21 Geth ¦ roast: review: https://github.com/perl6/roast/commit/5091bdd995
21:21 synopsebot RT#132216 [new]: https://rt.perl.org/Ticket/Display.html?id=132216 [UNI] [TESTNEEDED] 'a' coll 'A" not Same but More with disabled tertiary and primary levels
21:21 Geth ¦ roast: 47a21ea017 | (Zoffix Znet)++ | S32-str/Collation.t
21:21 Geth ¦ roast: Remove debug output from TAP
21:21 Geth ¦ roast: review: https://github.com/perl6/roast/commit/47a21ea017
21:22 AlexDaniel- “Now, GitHub will help potential first-time contributors discover issues labeled with 「help wanted」 or 「good first issue」”
21:23 AlexDaniel- three word tag… awesome
21:24 AlexDaniel- now what, mark issues with three tags at the same time? /o\ 「LHF」 「Hacktoberfest」 「good first issue」
21:25 teatime please tell me there is literally one called [THREE WORD TAG]
21:25 teatime if there is, Zoffix made it.  I am sure.
21:25 * AlexDaniel- doesn't get the joke
21:28 Geth ¦ roast: 061b040e2b | (Zoffix Znet)++ | 2 files
21:28 Geth ¦ roast: Move dir() stress test into stresstest
21:28 Geth ¦ roast:
21:28 Geth ¦ roast: Also bump number of iterations
21:28 Geth ¦ roast: review: https://github.com/perl6/roast/commit/061b040e2b
21:28 Geth ¦ rakudo/nom: de564a51b8 | (Zoffix Znet)++ | t/spectest.data
21:28 Geth ¦ rakudo/nom: Add S32-io/dir-stress.t to list of test files to run
21:28 Geth ¦ rakudo/nom: review: https://github.com/rakudo/rakudo/commit/de564a51b8
21:31 teatime AlexDaniel-: Three Word Tag is a Three Word Tag, like TLA
21:31 teatime (Three Letter Acronyum)
21:31 AlexDaniel- ah
21:31 Zoffix Some of the fudges in S32-io/dir.t look bogus.... "dirname is not yet absolute RT #124786" It's not meant to be. "dir() returns IO::Local RT #124785" we ain't got an IO::Local
21:31 synopsebot RT#124786 [new]: https://rt.perl.org/Ticket/Display.html?id=124786 [IO] S32-io/dir.t line:18 reason: 'dirname is not yet absolute'
21:31 synopsebot RT#124785 [new]: https://rt.perl.org/Ticket/Display.html?id=124785 [IO] S32-io/dir.t line:16 reason: 'entries are still IO::Path'
21:32 AlexDaniel- like YABA…
21:34 Geth ¦ roast: 73cb9e7ace | (Zoffix Znet)++ | S32-io/dir.t
21:34 Geth ¦ roast: Remove/Unfudge test for IO::Local
21:34 Geth ¦ roast:
21:34 Geth ¦ roast: We don't have this type no more.
21:34 Geth ¦ roast: Closes Rt#124785: https://rt.perl.org/Ticket/Display.html?id=124785
21:34 Geth ¦ roast: review: https://github.com/perl6/roast/commit/73cb9e7ace
21:35 Zoffix m: dir.head.say
21:35 camelia rakudo-moar 5d3ebc: OUTPUT: «".cpanm".IO␤»
21:35 Zoffix ... the test looks for $*CWD :S
21:36 Zoffix Ah, .dirname ok
21:47 Geth ¦ roast: 66fb7a1a3b | (Zoffix Znet)++ | S32-io/dir.t
21:47 Geth ¦ roast: Rewrite bogus-ish dir.head.dirname test
21:47 Geth ¦ roast:
21:47 Geth ¦ roast: The routine returns IO::Path objects which are composed of $!CWD
21:47 Geth ¦ roast: and $!path. These objects can be made cheaper by setting $!CWD to
21:47 Geth ¦ roast: the dir being dir()ed and $!path to the found stuff--without having
21:47 Geth ¦ roast: to resolve anything. For this reason, it doesn't make much sense
21:47 Geth ¦ roast: to require .dirname to be set to absolute paths.
21:47 Geth ¦ roast:
21:47 Geth ¦ roast: Change the test to instead test that the generated IO::Paths actually
21:47 Geth ¦ roast: point to proper filesystem entities.
21:47 Geth ¦ roast: review: https://github.com/perl6/roast/commit/66fb7a1a3b
21:48 bartolin pmurias: hmm, I don't think the nqp failure was caused by the BUILDPLAN changes. I reverted the commit from lizmat++ and still got that error. but if I change this line with the next one, all tests are clean: https://github.com/perl6/nqp/blob/e9bca388a7655b9cee0b4477eaccc241y4131b743/src/vm/jvm/runtime/org/perl6/nqp/sixmodel/reprs/P6bigintInstance.java#L10
21:48 yoleaux Zoffix: move this to stresstest https://github.com/perl6/roast/commit/807ed8137f
21:49 Zoffix yoleaux: just did it bruh
21:50 Zoffix oh haha
21:50 Zoffix m: IO::Local.new
21:50 camelia rakudo-moar 5d3ebc: OUTPUT: «IO::Local is disallowed in restricted setting␤  in sub restricted at src/RESTRICTED.setting line 1␤  in method new at src/RESTRICTED.setting line 32␤  in block <unit> at <tmp> line 1␤␤»
21:58 bartolin pmurias: I don't understand why that makes a difference, but when I printed 'field' around here I got 'private static final BigInteger org.perl6.nqp.sixmodel.reprs.P6bigintInstance.SMALLEST_UNBOXABLE': https://github.com/perl6/nqp/blob/e9bca388a7655b9cee0b4477eaccc2414131b743/src/vm/jvm/runtime/org/perl6/nqp/runtime/Ops.java#L2778
22:07 Zoffix ZOFLOP: t/spec/S17-supply/supplier-preserving.t
22:07 Zoffix Seems only one test in the file " No races in Supplier::Preserving".. and the flop suggests there might be
22:07 Zoffix Oh
22:07 Zoffix Non-zero wait status: 11
22:07 Zoffix Parse errors: Bad plan.  You planned 1 tests but ran 0.
22:07 Zoffix So it was a SEGV. No idea why
22:08 jnthn :S
22:08 yoleaux 20:19Z <AlexDaniel-> jnthn: Can we open rakudo/issues on github? See https://irclog.perlgeek.de/perl6-dev/2017-10-11#i_15289848 and the start of the discussion on https://irclog.perlgeek.de/perl6-dev/2017-10-11#i_15289680
22:08 Geth ¦ rakudo/nom: 50a674346b | (Zoffix Znet)++ | src/RESTRICTED.setting
22:08 Geth ¦ rakudo/nom: Remove trailing whitespace
22:08 Geth ¦ rakudo/nom: review: https://github.com/rakudo/rakudo/commit/50a674346b
22:08 Geth ¦ rakudo/nom: 4b8a0ef651 | (Zoffix Znet)++ | src/RESTRICTED.setting
22:08 Geth ¦ rakudo/nom: Remove classes we no longer have in core
22:08 Geth ¦ rakudo/nom: review: https://github.com/rakudo/rakudo/commit/4b8a0ef651
22:09 jnthn Zoffix: I fixed a couple of SEGV sources yesterday; dunno if the revision files were bumped since
22:09 Zoffix Yeah, they were.
22:10 jnthn Well, guess we know 2 bugs it isn't, at least...
22:10 Zoffix :)
22:11 Zoffix FWIW, reproed with `while t/fudgeandrun t/spec/S17-supply/supplier-preserving.t; do true; done` on about 10th run
22:11 Zoffix No idea how to debug SEGVs; maybe dogbert2 will be able to offer his expertise
22:11 MasterDuke jnthn: any idea what could cause multis to always go through find_best_dispatchee?
22:11 jnthn MasterDuke: The result not being cacheable
22:12 MasterDuke and why might that be the case? too many arguments to the method?
22:12 jnthn That can be because of a where clause, signature unpack, or because it was called with flattenned args
22:13 jnthn May be other reasons but I can't think of them off hand
22:13 jnthn Pretty sure the "too many arguments" thing can't happen any more
22:13 Zoffix Filed SEGV as https://rt.perl.org/Ticket/Display.html?id=132273
22:13 jnthn (The old multi cache had a fixed limit, I think I elimianted it)
22:14 timotimo i wonder if reverting my recent commits helps that bug any?
22:14 Zoffix I'm on HEAD
22:14 MasterDuke hm, this is INTERPOLATE after i broke it out into `proto (|), (Iterable:D \var, ...), (Callable:D \var, ...), (Mu:D \var, ...), (Mu:U \var, ...)`
22:14 Zoffix Oh reverting..
22:14 Zoffix No idea
22:14 MasterDuke and i don't believe it's ever called with flattened args
22:16 jnthn MasterDuke: Hmm, then not immediately sure, but I'd start by looking at if add_to_cache is being called for that multi, and if so, would look in MVMMultiCache.c or whatever it's called
22:17 MasterDuke will do, thanks
22:17 MasterDuke timotimo: i've seen that test flop before
22:18 MasterDuke add_to_cache is being called the exact same number of times as find_best_dispatchee
22:19 MasterDuke now off to see if there's anything interesting in MVMMultiCache.c...
22:27 Geth ¦ roast: 30e9d1624e | (Zoffix Znet)++ | fudgeandrun
22:27 Geth ¦ roast: Propagate exit code
22:27 Geth ¦ roast: review: https://github.com/perl6/roast/commit/30e9d1624e
22:30 Zoffix :| github changed again
22:30 Zoffix ZofBot: I don't like things that are different!
22:30 ZofBot Zoffix, com/rakudo/rakudo/blob/4de858a555fe0f6abfa813f0ff29e32edc159c2f/src/Perl6/ModuleLoader
22:30 Zoffix Though this reminds how much the black bar annoyed me when it changed, but now I don't even pay attention to it
22:30 Zoffix buggable: zen
22:30 buggable Zoffix, "A follower of the way has neither form nor shape, neither root nor trunk; nor dwelling place; like a fish leaping in the water."
22:31 MasterDuke hm, not sure how to interpret the output of turning MVM_MULTICACHE_DEBUG on
22:32 jnthn MasterDuke: That's likely not very interesting (it shows what's in the cache iirc, while I think what you're after is the resasons things don't get inserted)
22:33 MasterDuke ah, so likely have to stick some fprintfs in MVM_multi_cache_add?
22:38 jnthn Probably, yeah
22:42 MasterDuke lots of `if (!cs->is_interned) return cache_obj;` happened. not the exact number of add_to_cache/find_best_dispatchee call though
22:44 timotimo then check the callsite_try_intern code
22:44 timotimo it's got the logic for what is allowed to get inlined
22:44 timotimo er, interned
22:48 MasterDuke a bunch of has_flattening and some num_flags >= MVM_INTERN_ARITY_LIMIT
22:50 MasterDuke not yet sure how to match them up to the INTERPOLATE calls
22:50 MasterDuke afk for a bit, but thanks, making progress
22:53 jnthn Sleep time for me; 'night o/
22:54 Zoffix \o
22:54 timotimo gnite jnthn :)
22:54 timotimo i'll also go rest soon
23:12 MasterDuke ah, think it's arity
23:15 BenGoldberg joined #perl6-dev
23:17 timotimo hmpf, both JSON::Tiny and JSON::Fast generate invalid json when enum values are involved
23:17 timotimo c: use JSON::Fast; enum Blop <Moo Meep Squee>; say to-json([Moo, Meep, Squee]).perl;
23:17 committable6 timotimo, ¦use: «Cannot find this revision (did you mean “all”?)»
23:18 timotimo c: HEAD use JSON::Fast; enum Blop <Moo Meep Squee>; say to-json([Moo, Meep, Squee]).perl;
23:18 committable6 timotimo, https://gist.github.com/2d51be550475556f1f4f61a12b20cf7e
23:18 timotimo oh, not available, eh?
23:18 timotimo anyway, the output is just [ Moo, Meep, Squee ]  i.e. missing quotation marks
23:19 MasterDuke i can reproduce it manually. `multi method f(Mu:D \v, int $i, int $m, int $monkey, int $s, int $a)` is ok, but add one more parameter and no caching
23:19 timotimo oh? that's a somewhat low limit, isn't it?
23:19 MasterDuke src/core/callsite.h:106:#define MVM_INTERN_ARITY_LIMIT 8
23:20 timotimo yeah, and with it being a method it counts the $self, too
23:20 MasterDuke hm, comment right above it: /* Maximum arity + 1 that we'll intern callsites by. */
23:22 AlexDaniel- c: HEAD use lib ‘data/all-modules/cpan/TIMOTIMO/JSON-Fast’; use JSON::Fast; enum Blop <Moo Meep Squee>; say to-json([Moo, Meep, Squee]).perl;
23:22 committable6 AlexDaniel-, ¦HEAD(4b8a0ef): «"[\n  0,\n  1,\n  2\n]"»
23:22 AlexDaniel- timotimo: ↑
23:26 MasterDuke ugh, can't just increase MVM_INTERN_ARITY_LIMIT, that causes segfaults
23:31 timotimo oh?
23:34 timotimo i can't get it to output them as integer values
23:34 timotimo something to figure out tomorrow. gnite!
23:49 Zoffix .
23:49 yoleaux 23:07Z <HoboWithAShotgun> Zoffix: i saw your video ( https://www.youtube.com/watch?v=paa3niF72Nw ), that was very enjoyable. why not make a few more?
23:52 Zoffix haha: https://twitter.com/bcrypt/status/918222753500508160
23:52 Zoffix rakudo (nom)$ git log --all --pretty='%H %ae' | sort -k1,1 | head -1
23:53 Zoffix 00006b16dd961d20bc20e1ac1d32df6d44320e05 liz@dijkmat.nl
23:53 Zoffix :)
23:53 Zoffix cosimo for roast and smsl for docs :)
23:56 evalable6 joined #perl6-dev

| Channels | #perl6-dev index | Today | | Search | Google Search | Plain-Text | summary