Perl 6 - the future is here, just unevenly distributed

IRC log for #perl6-dev, 2017-11-02

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

All times shown according to UTC.

Time Nick Message
00:04 Geth ¦ rakudo: d21c31e1d7 | (Elizabeth Mattijsen)++ | lib/Telemetry.pm6
00:04 Geth ¦ rakudo: Add Telemetry msgsnd,msgrcv,nsignals,nvcsw,invcsw
00:04 Geth ¦ rakudo:
00:04 Geth ¦ rakudo: - Also add invcsw (involuntary context switches) to default report
00:04 Geth ¦ rakudo:   - this shows some data, at least on MacOS
00:04 Geth ¦ rakudo: - This concludes all the fields in nqp::getrusage (*phew*)
00:04 Geth ¦ rakudo: review: https://github.com/rakudo/rakudo/commit/d21c31e1d7
00:06 lizmat and that concludes my hacking for today
00:06 lizmat .tell gfldex hope this is enough to write a nice blogpost about Telemetry :-)
00:06 yoleaux lizmat: I'll pass your message to gfldex.
00:06 lizmat good night, #perl6-dev!
00:19 Zoffix \o
00:49 * Zoffix commits to remaking the entire Introducing Perl 6 brochure
00:50 Zoffix Took me half a year to realize I don't have to use $work's tiny laptop to work remotely and instead can remote-desktop into it from my desktop with giant monitor. So I now have proper hardware for design software when I'm hope :P
00:50 Zoffix *home
00:50 timotimo oh, nice
00:51 timotimo but what's wrong about inkscape!!! (lol omg)
00:52 Zoffix No idea. Never used it. Out of free tools, I've used Scribus 5 years ago and it kept crashing.
00:53 Zoffix Or making 1GB PDFs :)
00:53 timotimo fair enough
00:53 timotimo inkscape's UI has been in desperate need of a new UI for a long long time
00:53 timotimo there are people capable of doing spectacular things with it, of course
01:16 travis-ci joined #perl6-dev
01:16 travis-ci Rakudo build passed. Elizabeth Mattijsen '-Inf .. Inf should be NaN, not 0'
01:16 travis-ci https://travis-ci.org/rakudo/rakudo/builds/295829183 https://github.com/rakudo/rakudo/compare/5eeb72a9f514...21efe96ffb9d
01:16 travis-ci left #perl6-dev
01:42 samcv good * everybody
01:50 timotimo yo samcv
01:50 samcv hey
01:51 samcv timotimo, what do you think of a function for composing ropes. that we could use in concatenation and other functions. so you initialize an MVMstruct for the string, and the ncall a function to add strands onto it. where would that be a good place to use other than join and concatenation functions? anywhere else where that would be benefictial?
01:53 timotimo if you include repetitions, then the x operator maybe?
01:59 samcv well. we're not talking about the x operator
02:00 timotimo other than that i have no good ideas
02:02 samcv https://gist.github.com/samcv/5c2eb18fa5d4d428c081ab3e8bca8fc4 something like this
02:02 samcv a mvm function that would let us easily compose ropes
02:03 timotimo i can see the appeal
02:04 samcv one of the things i noticed is it's not trivial to compose ropes. we have concatenation but nothing else really does it. so i'm thinking if i have enough reasons for doing this
02:04 samcv it may be worth it to add these functions
02:07 samcv i'm basically asking someone to convince me i should do this :)
02:07 BenGoldberg joined #perl6-dev
02:12 timotimo the code looks good. it'll probably come out the same after gcc throws all its optimizations at it, but it's likely less error-prone
02:18 timotimo i'm going to bed, seeya!
02:34 Zoffix Man, I'm feeling the pain with all our Unicode ops and Unicode in outpout (e.g 「」). Even Noto font doesn't render them in InDesign; looks like whatever fallback mechanism usually exist doesn't work in typesetting software :(
02:37 Zoffix Finally managed to typeset this snippet with 3 fonts :P https://gist.github.com/zoffixznet/a4f23c4c647feaadbbe9ded22f514f5a
02:39 Zoffix Doesn't quite work still, 'cause I wanted to highlight it but can't apply two char styles together :|
02:57 ilbot3 joined #perl6-dev
02:57 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
06:24 travis-ci joined #perl6-dev
06:24 travis-ci Rakudo build passed. Elizabeth Mattijsen 'More Telemetry additions
06:24 travis-ci https://travis-ci.org/rakudo/rakudo/builds/295954168 https://github.com/rakudo/rakudo/compare/5e7dfe52fd6c...3b4f0c6ceb20
06:24 travis-ci left #perl6-dev
07:12 travis-ci joined #perl6-dev
07:12 travis-ci Rakudo build passed. Elizabeth Mattijsen 'Increase the size of gtc/ttc, gfldex++'
07:12 travis-ci https://travis-ci.org/rakudo/rakudo/builds/295979112 https://github.com/rakudo/rakudo/compare/7144dc290c48...73e1faaa713b
07:12 travis-ci left #perl6-dev
07:30 gfldex .
07:30 yoleaux 00:06Z <lizmat> gfldex: hope this is enough to write a nice blogpost about Telemetry :-)
07:32 gfldex .tell lizmat could you make %format available to members of the public please?
07:32 yoleaux gfldex: I'll pass your message to lizmat.
07:34 bartolin samcv: it looks like we have a regression with some tests for "\c[BELL]" on jvm. I think that might be related to your recent work on control characters (nqp commit 7954f9b0c1).
07:34 bartolin m: say "\c[BELL]".uniname  # gives <control-0007> on jvm
07:34 camelia rakudo-moar d21c31e1d: OUTPUT: «BELL␤»
07:35 bartolin samcv: I believe that happens because nqp::ord("\c[BELL]") returns 7 on jvm
07:36 bartolin samcv: do you have a pointer, perhaps, how "\c[BELL]" is processed?
07:40 bartolin samcv: I've read through RT #130542 and if I understood correctly, BELL is a special case, because it was U+0007 in Unicode 1. So, maybe we need to add a correcsponding special case somewhere. (I saw that you already did that for nqp::codepointfromname in nqp commit deb8cb03e.)
07:40 synopsebot RT#130542 [resolved]: https://rt.perl.org/Ticket/Display.html?id=130542 [JVM] [UNI] \c[BELL] returns the U+0007 control code not U+1F514 BELL
07:40 lizmat Files=1229, Tests=75811, 318 wallclock secs (14.79 usr  5.38 sys + 2198.71 cusr 212.47 csys = 2431.35 CPU)
07:40 yoleaux 07:32Z <gfldex> lizmat: could you make %format available to members of the public please?
07:41 lizmat gfldex: I see that as an implementation detail, what would you like to achieve by that ?
07:41 bartolin samcv: ah, one last thing: I've seen those failing tests in S32-str/parse-names.t
07:41 bartolin work &
08:00 gfldex lizmat: I would like to display the descriptions.
08:02 gfldex lizmat: and when %format is exported &report could get a :$format
08:08 samcv bartolin, oh there's failing tests?
08:17 bartolin samcv: here is a gist of runnning S32-str/parse-names.t with a current rakudo-j: https://gist.github.com/usev6/2e15b87350d481eb73905bfc3ae9f952
08:18 samcv bartolin, on jvm?
08:18 samcv ah ok
08:18 bartolin *nods*
08:18 bartolin the tests are fine on moar
08:18 samcv well. it's due to utf8-16
08:18 samcv err utf-18
08:19 samcv *16
08:19 bartolin *g*
08:20 bartolin those tests have been passing before, though (didn't bisect, but I thought it might happen since https://github.com/perl6/nqp/commit/7954f9b0c1
08:35 gfldex m: class T does Associative { method AT-KEY(\k) { 42 but role :: { method desc { "answer" } } } }; say T.new<foo>.desc
08:35 camelia rakudo-moar d21c31e1d: OUTPUT: «answer␤»
08:35 gfldex lizmat: maybe with a sneaky ↑↑↑ role?
09:04 Geth ¦ rakudo: 0dc4a0ebad | (Elizabeth Mattijsen)++ | lib/Telemetry.pm6
09:04 Geth ¦ rakudo: Telemetry tweaks #1 for today
09:04 Geth ¦ rakudo:
09:04 Geth ¦ rakudo: - handle 0 wallclock better, apparently Windows has bigger granularity
09:04 Geth ¦ rakudo: - add initial memory size (max-rss) to report
09:04 Geth ¦ rakudo: review: https://github.com/rakudo/rakudo/commit/0dc4a0ebad
09:19 lizmat m: multi sub a(:@c is copy) { }; a   # WOW, something we forgot in the GLR
09:19 camelia rakudo-moar 0dc4a0eba: OUTPUT: «replace this Array is copy logic␤  in sub a at <tmp> line 1␤  in block <unit> at <tmp> line 1␤␤»
09:21 lizmat https://github.com/rakudo/rakudo/blob/master/src/Perl6/Metamodel/BOOTSTRAP.nqp#L443
09:25 robertle joined #perl6-dev
09:27 jnthn Wow!
09:27 jnthn I see it's a highly used feature ;)
09:28 lizmat apparently  :-)
09:29 lizmat oddly enough, uncommmenting the following lines appears to make it work
09:30 * lizmat runs a spectest just in case
09:30 samcv hah wow
09:30 lizmat jnthn: do you see anything obviously wrong with the code ?
09:32 jnthn Not immediately
09:32 lizmat basically the same code as for :%h is copy
09:32 samcv lizmat, looks like more in Any.pm
09:32 samcv XXX GLR comments
09:32 lizmat yeah, those I know about  :-)
09:33 lizmat I think :-)
09:34 [TuxCM] joined #perl6-dev
09:37 lizmat m: sub a(:@c is copy) { }; a
09:37 camelia rakudo-moar 0dc4a0eba: ( no output )
09:38 lizmat jnthn: apparently only in the case of a multi sub
09:38 Geth ¦ rakudo: be1e2879f3 | (Elizabeth Mattijsen)++ | src/Perl6/Metamodel/BOOTSTRAP.nqp
09:38 Geth ¦ rakudo: Make multi sub(:@c is copy) { } work
09:38 Geth ¦ rakudo:
09:38 Geth ¦ rakudo: Apparently this was looked over during the GLR
09:38 Geth ¦ rakudo: review: https://github.com/rakudo/rakudo/commit/be1e2879f3
09:43 [TuxCM] Rakudo version 2017.10-95-g0dc4a0eba - MoarVM version 2017.10-29-g116c7a13
09:43 [TuxCM] csv-test-xs-20      0.423 -  0.465
09:43 [TuxCM] csv-ip5xs           1.105 -  1.109
09:43 [TuxCM] test-t              3.244 -  3.284
09:43 [TuxCM] test               11.907 - 12.084
09:43 [TuxCM] csv-parser         12.537 - 12.638
09:43 [TuxCM] csv-ip5xs-20       12.873 - 12.938
09:43 [TuxCM] test-t-20 --race   19.891 - 19.975
09:43 [TuxCM] test-t-20          58.383 - 58.828
09:47 Geth ¦ rakudo: c2baf95e1b | (Elizabeth Mattijsen)++ | lib/Telemetry.pm6
09:47 Geth ¦ rakudo: Telemetry tweaks #2 for today
09:47 Geth ¦ rakudo:
09:47 Geth ¦ rakudo: - "columns" is now a named parameter to "report"
09:47 Geth ¦ rakudo:   - specify the columns you want to see, either by header or method name
09:47 Geth ¦ rakudo: - %*ENV<RAKUDO_REPORT_COLUMNS> can now be used to set columns you want to see
09:47 Geth ¦ rakudo:   - if you don't specify any columns in the call to "report"
09:47 Geth ¦ rakudo: review: https://github.com/rakudo/rakudo/commit/c2baf95e1b
09:48 lizmat afk for a few hours&
10:00 travis-ci joined #perl6-dev
10:00 travis-ci Rakudo build passed. Elizabeth Mattijsen 'Add Telemetry is-rss,min-flt,maj-flt,nswap,inblock,outblock
10:00 travis-ci https://travis-ci.org/rakudo/rakudo/builds/296009383 https://github.com/rakudo/rakudo/compare/91e00e680174...96751ee87c93
10:00 travis-ci left #perl6-dev
10:18 |Tux| joined #perl6-dev
10:35 travis-ci joined #perl6-dev
10:35 travis-ci Rakudo build passed. Elizabeth Mattijsen 'Normalize all Telemetry x-rss to Kbytes
10:35 travis-ci https://travis-ci.org/rakudo/rakudo/builds/296013392 https://github.com/rakudo/rakudo/compare/96751ee87c93...7e00908cb272
10:35 travis-ci left #perl6-dev
10:45 astj joined #perl6-dev
11:29 bisectable6 joined #perl6-dev
11:33 travis-ci joined #perl6-dev
11:33 travis-ci Rakudo build failed. Elizabeth Mattijsen 'Add Telemetry msgsnd,msgrcv,nsignals,nvcsw,invcsw
11:33 travis-ci https://travis-ci.org/rakudo/rakudo/builds/296023846 https://github.com/rakudo/rakudo/compare/7e00908cb272...d21c31e1d723
11:33 buggable [travis build above] ✓ All failures are due to: GitHub connectivity (1 failure).
11:33 travis-ci left #perl6-dev
12:09 llfourn joined #perl6-dev
12:25 Zoffix .tell timotimo there looks to be a stale PR by you to mp6o: https://github.com/perl6/modules.perl6.org/pull/86/files IIRC the problem with it was that data fetcher doesn't fetch travis files
12:25 yoleaux Zoffix: I'll pass your message to timotimo.
12:26 pmurias joined #perl6-dev
12:27 travis-ci joined #perl6-dev
12:27 travis-ci Rakudo build passed. Elizabeth Mattijsen 'Telemetry tweaks #1 for today
12:27 travis-ci https://travis-ci.org/rakudo/rakudo/builds/296164052 https://github.com/rakudo/rakudo/compare/d21c31e1d723...0dc4a0ebad21
12:27 travis-ci left #perl6-dev
12:46 Zoffix Some feedback on our state: https://www.reddit.com/r/perl/comments/79wjtz/what_are_the_most_disliked_programming_languages/dp6r8j7/
12:49 lizmat m: class A { has Int @.a[4] = die }; dd A.new   # clearly the initializer is not being run
12:49 camelia rakudo-moar c2baf95e1: OUTPUT: «A.new(a => Array[Int].new(:shape(4,), [Int, Int, Int, Int]))␤»
12:50 lizmat m: class A { has Int @.a[4] is required }; dd A.new   # not even the required check ???
12:50 camelia rakudo-moar c2baf95e1: OUTPUT: «A.new(a => Array[Int].new(:shape(4,), [Int, Int, Int, Int]))␤»
12:50 pmurias Zoffix: re hard to navigate the docs search sucks
12:51 Zoffix pmurias: ok, good. That's one of the things that'll be improved by the redesign.
12:52 Zoffix That post actual points out a counter-point to people who say a new name for the language will nullify google results. It's actually a good thing in this case, where the user's saying they're getting outdated results :P
12:53 lizmat m: class A { has %.a is BagHash is required }; dd A.new  # looks like "is required" is a noop in these cases  :-(
12:53 camelia rakudo-moar c2baf95e1: OUTPUT: «A.new(a => ().BagHash)␤»
12:54 * Zoffix &
12:54 pmurias Zoffix: keeping the renaming debate in #perl6 out of #perl6-dev seems like a good idea ;)
12:54 Zoffix pmurias: why?
12:54 Zoffix To me, the reverse is true.
12:55 Zoffix The devs are the ones who should be the most involved in the discussion that concerns the marketing of this thing we're making.
12:57 Zoffix This is the boardroom with all the directors. You're making it seem like the suggestion to move the marketing discussions to the customer service deparment :)
12:57 * Zoffix leaves to do @urgent-things
13:11 Zoffix pmurias: oh and there's an easy way to reduce the amount of these discussions. Resolve the 6.d blocker so we could move forward with 6.d release: https://github.com/perl6/6.d-prep/blob/master/TODO/FEATURES.md#need-a-way-to-know-callers-language  The naming debate culminates then.
13:24 travis-ci joined #perl6-dev
13:24 travis-ci Rakudo build passed. Elizabeth Mattijsen 'Make multi sub(:@c is copy) { } work
13:24 travis-ci https://travis-ci.org/rakudo/rakudo/builds/296175854 https://github.com/rakudo/rakudo/compare/0dc4a0ebad21...be1e2879f37e
13:24 travis-ci left #perl6-dev
14:07 MasterDuke joined #perl6-dev
14:21 stmuk_ joined #perl6-dev
14:22 travis-ci joined #perl6-dev
14:22 travis-ci Rakudo build passed. Elizabeth Mattijsen 'Telemetry tweaks #2 for today
14:22 travis-ci https://travis-ci.org/rakudo/rakudo/builds/296179262 https://github.com/rakudo/rakudo/compare/be1e2879f37e...c2baf95e1b37
14:22 travis-ci left #perl6-dev
14:39 perlpilot joined #perl6-dev
15:10 Geth ¦ rakudo: 9f54bc989d | (Elizabeth Mattijsen)++ | src/core/Exception.pm
15:10 Geth ¦ rakudo: Add X::NYI.workaround
15:10 Geth ¦ rakudo:
15:10 Geth ¦ rakudo: So we can give workaround information if necessary
15:10 Geth ¦ rakudo: review: https://github.com/rakudo/rakudo/commit/9f54bc989d
15:21 japhb joined #perl6-dev
15:31 Geth ¦ rakudo: 0973b307ad | (Elizabeth Mattijsen)++ | src/Perl6/Metamodel/BUILDPLAN.nqp
15:31 Geth ¦ rakudo: Temporary fix for GH #1226
15:31 Geth ¦ rakudo:
15:31 Geth ¦ rakudo: - throw an X::Comp::NYI for buildplan code 9 with a 4,5,6,7
15:31 Geth ¦ rakudo: - show a work around
15:31 Geth ¦ rakudo:
15:31 Geth ¦ rakudo: Can't help but wonder that if we recommend this workaround, why not
15:31 synopsebot RAKUDO#1226 [open]: https://github.com/rakudo/rakudo/issues/1226 Broken default value assignment on shaped array attributes
15:31 Geth ¦ rakudo: automatically generate that code then?  This could be done by adding
15:31 Geth ¦ rakudo: another code block like TWEAK to the BUILDPLAN, with the correct
15:31 Geth ¦ rakudo: code pre-generated.
15:31 Geth ¦ rakudo: review: https://github.com/rakudo/rakudo/commit/0973b307ad
15:43 Zoffix m: class F { has @.color[4]; submethod TWEAK (:@!color = <1 2 3 4>) {} }; F.new.color.say; # shorter FWIW
15:43 camelia rakudo-moar 9f54bc989: OUTPUT: «[1 2 3 4]␤»
15:44 lizmat m: class F { has @.color[4]; submethod TWEAK (:@!color = <1 2 3 4>) {} }; F.new(color => <f g h i>).color.say;
15:44 camelia rakudo-moar 9f54bc989: OUTPUT: «[f g h i]␤»
15:45 lizmat ah, good point !
15:52 Geth ¦ rakudo: 6dab5aad89 | (Elizabeth Mattijsen)++ | src/Perl6/Metamodel/BUILDPLAN.nqp
15:52 Geth ¦ rakudo: Improve GH #1226 workaround suggestion, Zoffix++
15:52 Geth ¦ rakudo: review: https://github.com/rakudo/rakudo/commit/6dab5aad89
15:52 synopsebot RAKUDO#1226 [open]: https://github.com/rakudo/rakudo/issues/1226 Broken default value assignment on shaped array attributes
15:54 * Zoffix wonders if that'll make it come out as `(:@.color …` instead of @! twigil
15:55 lizmat 'class F { has @.color[4] = <1 2 3 4> }' comes out as 'method TWEAK(:@!color = (initial values)) { }' for me
15:56 Zoffix \o/
15:57 hankache joined #perl6-dev
15:59 lizmat m: use Telemetry; say T<max-rss>
15:59 camelia rakudo-moar 0973b307a: OUTPUT: «91504␤»
15:59 lizmat m: use Telemetry; say T<max-rss>
15:59 camelia rakudo-moar 0973b307a: OUTPUT: «87740␤»
15:59 lizmat wonder where these differences come from
16:00 lizmat m: use Telemetry :COLUMNS; say max-rss   # wow
16:00 camelia rakudo-moar 0973b307a: OUTPUT: «116860␤»
16:00 ugexe what does it mean if there is no max-rss?
16:00 yoleaux 08:54Z <tbrowder> ugexe: how can i get a local install of rakudo to find Debugger::UI::CommandLine (and its dependencies) in system-wide COMPUNITS? alternatively, how can i get zef to —install-to=/some/path?
16:01 lizmat ugexe: that your OS doesn't support it
16:01 lizmat http://man7.org/linux/man-pages/man2/getrusage.2.html
16:01 ugexe lizmat: on osx for 1..10 { snap } shows max-rss twice but the rest are blank
16:02 lizmat ah, blank means no change, it hides 0 as that makes for a very noisy overview
16:02 ugexe i see, same for ics/gw/etc?
16:02 lizmat yup
16:03 lizmat otherwise you're just looking at a page with 0's
16:03 lizmat and some none zero values
16:03 Zoffix tbrowder: what I usually do is run Step 12 from the release guide except for the last `zef install ...` replaced with the module I need: https://github.com/rakudo/rakudo/blob/nom/docs/release_guide.pod
16:07 tbrowder Zoffix: thanks! BTW, any updates to yr markdown to html for advent entries since last year?
16:10 Zoffix Nope. I'm not sure where the code is at even. Probably can be improved a bit and shared with all the authors.
16:15 astj joined #perl6-dev
16:21 Zoffix So yesterday I was whinging because there wasn't an easy way to style code parts in InDesign to have syntax highlights. Turns out it supports PCRE in paragraph styles to apply substyles. And a bunch of regex hackery later, I managed to make a single paragraph style that reasonably-well highlights Perl 6 code up in it: https://i.imgur.com/1GsZsBZ.png
16:21 Zoffix Not that anyone here cares about InDesign, but it'll make creating marketing brochures with code about 1000x faster :)
16:22 Zoffix ZofBot: do the happy dance, bud
16:22 ZofBot Zoffix, 'Would you have done all that for me?' she said
16:22 Zoffix For you, I would've!
16:23 unicodable6 joined #perl6-dev
16:24 timotimo the other day i was trying something with List
16:25 timotimo the reason why i even started trying the nqp::attrinited stuff on List was that $!todo.DEFINITE will autovivify a Scalar container even for a List object that'll never have (and never had) an iterator part
16:26 timotimo so this is perhaps an opportunity for improvement. though maybe we actually want to do this in the optimizer
16:38 lizmat hmmm... good one
16:38 timotimo okay, so ...
16:38 timotimo i outputted every "demanded" object and stable with Telemetry vs Telemetry :COLUMNS
16:39 timotimo and with :COLUMNS it's actually fewer things in there
16:39 timotimo it could be that more stuff hangs off of these, though
16:44 timotimo nope, even with the work loop itself instrumented it's fewer things if :COLUMNS is specified
16:44 timotimo time for the heap analyzer
16:48 timotimo nothing very obvious
17:05 Geth ¦ rakudo: e1a1b8d865 | (Elizabeth Mattijsen)++ | lib/Telemetry.pm6
17:05 Geth ¦ rakudo: Add support for Telemetry affinity-tasks-completed
17:05 Geth ¦ rakudo:
17:05 Geth ¦ rakudo: - also add this to the default report
17:05 Geth ¦ rakudo: review: https://github.com/rakudo/rakudo/commit/e1a1b8d865
17:14 jnthn Urgh, a bug I've been hunting for the last few hours might boil down to something wrong with the CLOSE phaser
17:16 lizmat whee
17:17 lizmat also: if you can't start another thread, should we act on that somehow more than just ignoring ?
17:18 jnthn In what sense?
17:18 jnthn Thread.start explodes pretty forcefully, and perhaps too much so, if it fails
17:19 jnthn (It's a VM panic)
17:19 lizmat m: await start { sleep 1 } xx 64; say T.general-workers  # never goes beyond 47 for me
17:19 camelia rakudo-moar e1a1b8d86: OUTPUT: «5===SORRY!5=== Error while compiling <tmp>␤Undeclared name:␤    T used at line 1␤␤»
17:19 lizmat m: use Telemetry; await start { sleep 1 } xx 64; say T.general-workers  # never goes beyond 47 for me
17:19 camelia rakudo-moar e1a1b8d86: OUTPUT: «MoarVM panic: Could not spawn thread: errorcode -11␤»
17:19 lizmat ooh wow
17:19 jnthn That's 'cus of limits in the environment camelia runs under
17:19 lizmat on MacOS it just says 47
17:20 lizmat if run with RAKUDO_SCHEDULER_DEBUG, I get a lot of: Heuristic queue progress deadlock situation detected
17:20 lizmat followed by Could not create a new Thread: too many open files
17:20 jnthn Ah
17:20 jnthn Well, I think it's probably doing the right thing
17:21 jnthn No reason to bring down a program that will eventually work
17:21 jnthn Instead of leaving it to complete
17:24 Zoffix I still get "too many open files" even if I set RAKUDO_MAX_THREADS=10
17:25 ilmari what does 'ulimit -n' output?
17:26 Zoffix hm, nm, setting it to 7 does go through the "max threads reached" message.
17:26 perlpilot "too many open files" is a strange way to say ENOMORETHREADS
17:26 Zoffix ilmari: I set it to 50
17:26 Zoffix But I guess there are other things using files, so the left over is 7 for threads :)
17:26 jnthn perlpilot: Probably really means "handles"
17:29 tbrowder Zoffix: I have a copy of your md2html from last year. It's at github gist: https://gist.github.com/tbrowder/32154ec024d1d36003e84358fd4823d8, plus my code wrapping it for my md source. Both are at gist
17:30 tbrowder ignore words after the gist, please
17:33 Zoffix m: await start { sleep 1 } xx 30
17:33 camelia rakudo-moar e1a1b8d86: OUTPUT: «MoarVM panic: Could not spawn thread: errorcode -11␤»
17:33 Zoffix So what limitations does camelia's env have? I set all the same ulimit values (a few days ago) and didn't have that issue locally.
17:35 Zoffix It's kind'f an unwanted issue IMO. The work can be done on a couple of threads really and the program I wrote that works elsewhere crashes here for reasons I don't know.
17:35 Zoffix (this also didn't exist in 2017.08)
17:35 Zoffix *.07
17:35 Zoffix star: await start { sleep 1 } xx 30
17:35 camelia star-m 2017.07: ( no output )
17:36 jnthn Then set RAKUDO_MAX_THREADS=16 and you'll have the 2017.07 max pool size back
17:36 Zoffix m: BEGIN %*ENV<RAKUDO_MAX_THREADS> = 16; await start { sleep 1 } xx 30
17:36 camelia rakudo-moar e1a1b8d86: OUTPUT: «MoarVM panic: Could not spawn thread: errorcode -11␤»
17:37 jnthn I think it actually has to be in the env
17:37 Zoffix m: BEGIN %*ENV<RAKUDO_MAX_THREADS> = 2; await start { sleep 1 } xx 30
17:37 camelia rakudo-moar e1a1b8d86: OUTPUT: «MoarVM panic: Could not spawn thread: errorcode -11␤»
17:37 jnthn As the scheduler instance is created at CORE.setting load time
17:37 Zoffix 2 just worked in #zofbot
17:38 jnthn Can perhaps also look into whether the panic can be turned into a throw
17:38 jnthn In which case the sueprvisor will be able to silently swallow it
17:38 Zoffix m: use nqp; nqp::bindattr(nqp::decont($*SCHEDULER), ThreadPoolScheduler, q|$!max_threads|, 16);  await start { sleep 1 } xx 30
17:38 jnthn Just needs some care to make sure that we don't end up with a corrupt internal VM state as a result
17:38 camelia rakudo-moar e1a1b8d86: ( no output )
17:38 Zoffix k, yeah, works with 16
17:39 jnthn Which may be why it's a panic now
17:40 jnthn Looking very much like the CLOSE phaser bug was the root cause of the problem I was hunting. Thankfully this case is trivial to rewrite in terms of .on-close, which is safe
17:41 jnthn I'll file an issue later, or mebbe tomorrow
17:48 jnthn Phew. That was some debugging. But, we'll get Cro 0.7.1 this week (or weekend) thanks to it. :)
17:49 * jnthn disappers to cook pau bahji
18:00 Geth ¦ rakudo: b380230dd9 | (Elizabeth Mattijsen)++ | lib/Telemetry.pm6
18:00 Geth ¦ rakudo: Add support for Telemetry affinity-tasks-queued
18:00 Geth ¦ rakudo: review: https://github.com/rakudo/rakudo/commit/b380230dd9
18:01 * lizmat steps away for a few hours to grok
18:09 Ven joined #perl6-dev
18:50 Zoffix ZofBot: twerk it: https://www.youtube.com/watch?v=-pqvQ8k308I
18:50 ZofBot Zoffix, It's a hobby of mine
18:50 * Zoffix rofls
18:59 MasterDuke joined #perl6-dev
19:06 buggable New CPAN upload: bamboo-0.000.003_20171102.tar.gz by PSIXDISTS https://cpan.metacpan.org/authors/id/P/PS/PSIXDISTS/Perl6/bamboo-0.000.003_20171102.tar.gz
19:06 buggable New CPAN upload: DateTime-Parse-0.000.003_20171102.tar.gz by PSIXDISTS https://cpan.metacpan.org/authors/id/P/PS/PSIXDISTS/Perl6/DateTime-Parse-0.000.003_20171102.tar.gz
19:07 ugexe Supply has no .close, but there was a multi sub close(Supply:D) added
19:09 [TuxCM] joined #perl6-dev
19:11 Geth ¦ rakudo: ef84aafc01 | (Nick Logan)++ (committed using GitHub Web editor) | src/core/io_operators.pm
19:11 Geth ¦ rakudo: Remove invalid close() candidate
19:11 Geth ¦ rakudo:
19:11 Geth ¦ rakudo: Supply has no close method
19:11 Geth ¦ rakudo: review: https://github.com/rakudo/rakudo/commit/ef84aafc01
19:32 Geth ¦ roast: 02f72506a6 | usev6++ | 4 files
19:32 Geth ¦ roast: [jvm] Unfudge some passing tests
19:32 Geth ¦ roast: review: https://github.com/perl6/roast/commit/02f72506a6
19:33 [Coke] ... does the cpan upload of bamboo also depend on panda?
19:33 Ven joined #perl6-dev
19:33 Zoffix I don't know why the PSIXDISTS user is uploading anything. It's ignored by modules.perl6.org
19:34 Zoffix I thought we were doing those uploads only as an experimental measure. They're violating the license of some modules.
19:39 lizmat mst should know more
19:39 lizmat but I was also under the impression that PSIXDIST was hated
19:39 lizmat *halted
19:39 lizmat oops
19:39 Zoffix :P
19:40 mst I'll make sure the cron job gets turned off
19:40 mst it was really just a way to remind perl6 that cpan actually existed
19:40 lizmat ugexe: good catch: I just acked for "method close", saw there was one in Supply.pm and added it
19:41 mst now you appear to have finally noticed it can go away
19:41 lizmat afk again&
19:41 Zoffix mst++
19:42 Zoffix mst: FWIW, I'm including this blurb about Perl in a new Rakudo Perl 6 brochure: https://gist.github.com/zoffixznet/b8e883c31e9c969405f8cfdfbe2ed17a
19:42 Zoffix If it needs to be tweaked, let me know.
19:43 mst looks fine to me
19:43 Zoffix \o/
19:47 AlexDaniel joined #perl6-dev
20:12 [Coke] Zoffix: yup, looks good.
20:14 jnthn Yup, it's Tap that has close, not Supply
20:18 perlpilot Zoffix: check the comments on that gist
20:18 perlpilot Zoffix++ btw
20:31 astj joined #perl6-dev
20:36 mst Zoffix: cronjob is now off.
20:49 Zoffix Sweet. Thanks.
21:24 Geth ¦ nqp: usev6++ created pull request #380: [jvm] Fix \c[BELL] returning U+0007
21:24 Geth ¦ nqp: review: https://github.com/perl6/nqp/pull/380
21:24 synopsebot NQP#380 [open]: https://github.com/perl6/nqp/pull/380 [jvm] Fix \c[BELL] returning U+0007
21:27 bartolin samcv: ^^ if you have a minute or two to look at that PR, that would be great
21:55 samcv thanks bartolin will look in a sec
21:55 samcv bartolin, i thought i fixed that previously hmm
21:56 samcv oh. awesome thanks bartolin
21:58 Geth ¦ nqp: 66e58e8f59 | usev6++ | src/vm/jvm/runtime/org/perl6/nqp/runtime/Ops.java
21:58 Geth ¦ nqp: [jvm] Fix \c[BELL] returning U+0007
21:58 Geth ¦ nqp: review: https://github.com/perl6/nqp/commit/66e58e8f59
21:58 Geth ¦ nqp: 5a03f1e9c6 | (Samantha McVey)++ (committed using GitHub Web editor) | src/vm/jvm/runtime/org/perl6/nqp/runtime/Ops.java
21:58 Geth ¦ nqp: Merge pull request #380 from usev6/jvm_bell
21:58 Geth ¦ nqp:
21:58 Geth ¦ nqp: [jvm] Fix \c[BELL] returning U+0007
21:58 synopsebot NQP#380 [closed]: https://github.com/perl6/nqp/pull/380 [jvm] Fix \c[BELL] returning U+0007
21:58 Geth ¦ nqp: review: https://github.com/perl6/nqp/commit/5a03f1e9c6
22:00 bartolin \o/ thanks samcv++ # all that work on unicode stuff
22:05 samcv you're welcome :-)
22:10 Geth ¦ nqp: de0f955305 | (Samantha McVey)++ | tools/build/MOAR_REVISION
22:10 Geth ¦ nqp: Bump MoarVM for join and Unicode property improvements
22:10 Geth ¦ nqp:
22:10 Geth ¦ nqp: Full changelog:
22:10 Geth ¦ nqp: e76788819 unicode: Make sure 'Cn' codepoints match with C enum
22:10 Geth ¦ nqp: 25bbba8c5 unicode: make sure to return 0 not "" from get_property_int
22:10 Geth ¦ nqp: e4b39aa1d show the actual type when "no such attribute" error occurs
22:10 Geth ¦ nqp: 87191078c Add MVMROOT2..MVMROOT5 to make rooting multiple variables easier
22:10 Geth ¦ nqp: 4136ae256 speed up join of longer strings by a significant amount
22:10 Geth ¦ nqp: 11ab1dcbe join: factor code into join_get_str_from_pos() function
22:10 Geth ¦ nqp: review: https://github.com/perl6/nqp/commit/de0f955305
22:10 Geth ¦ nqp: version bump brought these changes: https://github.com/MoarVM/MoarVM/compare/2017.10-29-g116c7a133...2017.10-37-ge76788819
22:21 cognominal joined #perl6-dev
22:30 travis-ci joined #perl6-dev
22:30 travis-ci NQP build failed. Samantha McVey 'Bump MoarVM for join and Unicode property improvements
22:30 travis-ci https://travis-ci.org/perl6/nqp/builds/296505978 https://github.com/perl6/nqp/compare/5a03f1e9c6ee...de0f955305b3
22:30 travis-ci left #perl6-dev
22:58 Zoffix .ask lizmat are you able to be to advise on the brochure or have access to woolfy to ask? I see it has grown considerably in size, driving up printing costs. I don't think my version will be any larger than the current 60-page one, but should I compress it as much as I can to keep the costs down or are the costs less important than more pleasant aesthetics?
22:58 yoleaux Zoffix: I'll pass your message to lizmat.
22:59 woolfy joined #perl6-dev
23:05 lizmat .
23:05 yoleaux 22:58Z <Zoffix> lizmat: are you able to be to advise on the brochure or have access to woolfy to ask? I see it has grown considerably in size, driving up printing costs. I don't think my version will be any larger than the current 60-page one, but should I compress it as much as I can to keep the costs down or are the costs less important than more pleasant aesthetics?
23:05 lizmat :-)
23:27 evalable6 joined #perl6-dev
23:40 Geth ¦ nqp: 8e0cd7a4e8 | MasterDuke17++ | src/vm/moar/QAST/QASTOperationsMAST.nqp
23:40 Geth ¦ nqp: Add the new coerce_II op as fromI_I
23:40 Geth ¦ nqp: review: https://github.com/perl6/nqp/commit/8e0cd7a4e8
23:40 Geth ¦ nqp: b8c85f824b | (Zoffix Znet)++ (committed using GitHub Web editor) | src/vm/moar/QAST/QASTOperationsMAST.nqp
23:40 Geth ¦ nqp: Merge pull request #378 from MasterDuke17/add_coerce_II_op
23:40 Geth ¦ nqp:
23:40 Geth ¦ nqp: Add the new coerce_II op as fromI_I
23:40 synopsebot NQP#378 [closed]: https://github.com/perl6/nqp/pull/378 Add the new coerce_II op as fromI_I
23:40 Geth ¦ nqp: review: https://github.com/perl6/nqp/commit/b8c85f824b
23:46 jnthn Zoffix: Just in case you forgot: that NQP merge'll need a MoarVM version bump
23:50 Zoffix yup
23:54 samcv i bumped MVM already
23:54 samcv oh looks like new commits. nvm
23:54 Zoffix TODO passed:   543-545, 549 t/spec/S05-mass/properties-general.t
23:54 Zoffix ZOFFLOP: t/spec/S11-modules/nested.t t/spec/S11-modules/require.t
23:55 Zoffix ZOFVM: Files=1283, Tests=152820, 155 wallclock secs (21.58 usr  3.17 sys + 3322.77 cusr 193.19 csys = 3540.71 CPU)
23:55 Geth ¦ nqp: a3f7bb8177 | (Zoffix Znet)++ | tools/build/MOAR_REVISION
23:55 Geth ¦ nqp: Bump MoarVM
23:55 Geth ¦ nqp: review: https://github.com/perl6/nqp/commit/a3f7bb8177
23:55 Geth ¦ nqp: version bump brought these changes: https://github.com/MoarVM/MoarVM/compare/2017.10-37-ge76788819...2017.10-47-gf88ab16
23:55 Geth ¦ rakudo: dbf0a21f90 | (Zoffix Znet)++ | tools/build/NQP_REVISION
23:55 Geth ¦ rakudo: Bump NQP
23:55 Geth ¦ rakudo: review: https://github.com/rakudo/rakudo/commit/dbf0a21f90
23:56 travis-ci joined #perl6-dev
23:56 travis-ci NQP build failed. Zoffix Znet 'Merge pull request #378 from MasterDuke17/add_coerce_II_op
23:56 travis-ci https://travis-ci.org/perl6/nqp/builds/296540731 https://github.com/perl6/nqp/compare/de0f955305b3...b8c85f824b30
23:56 travis-ci left #perl6-dev
23:56 Geth ¦ rakudo: 6cb7ebfb9f | MasterDuke17++ | src/core/Int.pm
23:56 Geth ¦ rakudo: Use the new nqp::fromI_I in Int.new
23:56 Geth ¦ rakudo:
23:56 Geth ¦ rakudo: The op will create a new Int using the value of the given Int, but
23:56 Geth ¦ rakudo: remove any mix-ins.
23:56 Geth ¦ rakudo: review: https://github.com/rakudo/rakudo/commit/6cb7ebfb9f
23:56 Geth ¦ rakudo: 264eef7988 | (Zoffix Znet)++ (committed using GitHub Web editor) | src/core/Int.pm
23:56 samcv Zoffix, you can untodo those now passing tests now that rakudo has been bumped if you want Zoffix. otherwise i'll get to it a little later
23:56 Geth ¦ rakudo: Merge pull request #1217 from MasterDuke17/use_fromI_I_in_Int.new
23:56 synopsebot RAKUDO#1217 [closed]: https://github.com/rakudo/rakudo/pull/1217 Use the new nqp::fromI_I in Int.new
23:56 Geth ¦ rakudo:
23:56 Geth ¦ rakudo: Use the new nqp::fromI_I in Int.new
23:56 Geth ¦ rakudo: review: https://github.com/rakudo/rakudo/commit/264eef7988
23:56 samcv err actually wait. gotta push some changes
23:58 Geth ¦ roast: b42fcb389c | (Zoffix Znet)++ | S05-mass/properties-general.t
23:58 Geth ¦ roast: Unfudge passing todos
23:58 Geth ¦ roast: review: https://github.com/perl6/roast/commit/b42fcb389c

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