Perl 6 - the future is here, just unevenly distributed

IRC log for #perl6-dev, 2017-11-03

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

All times shown according to UTC.

Time Nick Message
00:02 Geth ¦ roast: e3d5196640 | (Samantha McVey)++ | 2 files
00:02 Geth ¦ roast: Add additional tests for Unassigned characters (Cn) and <:C>
00:02 Geth ¦ roast:
00:02 Geth ¦ roast: Also untodo the now passing tests. These tests are for RT#124863
00:02 Geth ¦ roast: review: https://github.com/perl6/roast/commit/e3d5196640
00:02 synopsebot RT#124863 [open]: https://rt.perl.org/Ticket/Display.html?id=124863 [UNI] Roast rakudo skip/todo test:./S05-mass/properties-general.t line:735 reason: Unions of properties of non-existent codepoints
00:04 Zoffix Missed the "wait" :)
00:05 Zoffix All tests successful.
00:05 Zoffix Files=1283, Tests=152832, 153 wallclock secs (21.64 usr  3.23 sys + 3282.18 cusr 167.58 csys = 3474.63 CPU)
00:05 Zoffix Result: PASS
00:09 travis-ci joined #perl6-dev
00:09 travis-ci NQP build failed. Zoffix Znet 'Bump MoarVM'
00:09 travis-ci https://travis-ci.org/perl6/nqp/builds/296545753 https://github.com/perl6/nqp/compare/b8c85f824b30...a3f7bb81776a
00:09 travis-ci left #perl6-dev
00:15 Zoffix jvm stuff fails
00:15 Zoffix Oh, I guess it's missing the nqp::boxI_I op?
00:15 Zoffix Dunno; fails in t/nqp/014-while.t  and t/nqp/044-try-catch.t
00:18 jnthn sleep &
00:27 samcv o/ jnthn
00:30 BenGoldberg joined #perl6-dev
00:31 lizmat joined #perl6-dev
00:49 Geth ¦ nqp: 31844c5f88 | MasterDuke17++ | 2 files
00:49 Geth ¦ nqp: Implement nqp::fromI_I for JVM
00:49 Geth ¦ nqp: review: https://github.com/perl6/nqp/commit/31844c5f88
00:59 evalable6 joined #perl6-dev
01:13 travis-ci joined #perl6-dev
01:13 travis-ci NQP build failed. MasterDuke17 'Implement nqp::fromI_I for JVM'
01:13 travis-ci https://travis-ci.org/perl6/nqp/builds/296563077 https://github.com/perl6/nqp/compare/a3f7bb81776a...31844c5f88d4
01:13 travis-ci left #perl6-dev
01:18 Zoffix t/nqp/014-while.t  and t/nqp/044-try-catch.t
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
03:27 evalable6 joined #perl6-dev
04:15 MasterDuke nqp-m: my $x := 0; repeat until $x >= 2 -> $another_x {if $x == 0 { nqp::say(nqp::isnull($another_x)) } else { nqp::say($another_x) }; $x := $x + 1; }
04:15 camelia nqp-moarvm: OUTPUT: «1␤0␤»
04:15 MasterDuke nqp-j: my $x := 0; repeat until $x >= 2 -> $another_x {if $x == 0 { nqp::say(nqp::isnull($another_x)) } else { nqp::say($another_x) }; $x := $x + 1; }
04:16 MasterDuke well, that prints 0, 0 for nqp-j (t/nqp/014-while.t)
04:25 awwaiid joined #perl6-dev
06:19 evalable6 joined #perl6-dev
06:31 Geth ¦ nqp: 72487f7d81 | usev6++ | 2 files
06:31 Geth ¦ nqp: [jvm] Skip two newly added tests that fail on JVM
06:31 Geth ¦ nqp: review: https://github.com/perl6/nqp/commit/72487f7d81
06:58 travis-ci joined #perl6-dev
06:58 travis-ci NQP build failed. usev6 '[jvm] Skip two newly added tests that fail on JVM'
06:58 travis-ci https://travis-ci.org/perl6/nqp/builds/296639149 https://github.com/perl6/nqp/compare/31844c5f88d4...72487f7d81fa
06:58 travis-ci left #perl6-dev
07:02 bartolin hmm, looks like the travis job for jvm didn't run to completion ^^
07:07 bartolin how does one bump nqp? does it suffice to update rakudo's tools/build/NQP_REVISION with the output of 'git describe' (from the nqp repo)?
07:14 Geth ¦ nqp: 5f9e354ccf | (Christian Bartolomäus)++ (committed using GitHub Web editor) | t/nqp/014-while.t
07:14 Geth ¦ nqp: [jvm] Fix skip command (we skip two tests here)
07:14 Geth ¦ nqp: review: https://github.com/perl6/nqp/commit/5f9e354ccf
07:20 evalable6 joined #perl6-dev
07:31 travis-ci joined #perl6-dev
07:31 travis-ci NQP build passed. Christian Bartolomäus '[jvm] Fix skip command (we skip two tests here)'
07:31 travis-ci https://travis-ci.org/perl6/nqp/builds/296648915 https://github.com/perl6/nqp/compare/72487f7d81fa...5f9e354ccfcf
07:31 travis-ci left #perl6-dev
07:32 bartolin \o/
07:37 [Tux] Rakudo version 2017.10-106-g264eef798 - MoarVM version 2017.10-47-gf88ab169d
07:37 [Tux] csv-test-xs-20      0.438 -  0.484
07:37 [Tux] csv-ip5xs           1.113 -  1.129
07:37 [Tux] test-t              3.307 -  3.556
07:37 [Tux] test               12.043 - 12.147
07:37 [Tux] csv-parser         12.386 - 12.520
07:37 [Tux] csv-ip5xs-20       13.179 - 13.247
07:37 [Tux] test-t-20 --race   21.258 - 21.820
07:37 [Tux] test-t-20          60.880 - 61.919
08:49 pmurias joined #perl6-dev
08:50 pmurias The hint when getting attributes on moarvm and jvm isn't always correct?
09:05 patrickz joined #perl6-dev
09:53 Ven joined #perl6-dev
10:06 jnthn pmurias: It's invalid in the presence of MI
10:09 pmurias jnthn: the get attribute code is expected to verify if the hint is correct?
10:13 jnthn It's expected to ignore it in the case of MI; see the MoarVM impl where that's probably clearest
10:20 Zoffix bartolin: yeah, that's the way to do it. FWIW, here are my bumping bash aliases: https://gist.github.com/zoffixznet/8274ecdb83763b2509bca972d496c604
10:20 pmurias jnthn: thank you, I see it now
10:21 Zoffix (the second one is for manual push for when stresstest flops)
10:37 patrickz Hey! I'm trying to get my first doc commit ready. I'm struggling with `=begin code` blocks, which seem to indent further with every line in them (somehow even spanning multiple blocks).
10:37 patrickz This is how the code looks: https://gist.github.com/patzim/4b6d2c83132eb9f1b038d59f828904bf
10:38 patrickz This is the result: https://imgur.com/a/lXLGi
10:38 patrickz Any ideas?
10:39 bartolin Zoffix: thanks!
10:46 Zoffix patrickz: unsure what you mean. Got a screenshot/markup pastebin of problematic rendering?
10:49 patrickz Zoffix: https://imgur.com/a/lXLGi < I added a second image with the spaces I'm worried about marked.
10:52 Zoffix patrickz: and what's the resultant markup?
10:52 Zoffix That's not the render I get and I don't see how you end up with your version with the markup I get
10:53 patrickz Ah, the images are the html/perl6.xhtml file opened directly in firefox.
10:54 patrickz Is that at least somehow representative how it will end up on the website?
10:58 Zoffix patrickz: then I would ignore it and just commit and see what it comes out as on the website. html/perl6.xhtml has some escaping bugs ( https://github.com/perl6/doc/issues/1285 ) so it wouldn't surprise me if it also generates invalid markup that causes your issue
10:59 Zoffix patrickz: personally, I just push docs as is and don't worry much about rendering. It can be fixed later, if needed and this method lets me write all the docs quicker, without spending time to render them.
10:59 * Zoffix &
10:59 patrickz Will do. Thanks!
11:05 Ven joined #perl6-dev
11:07 woolfy left #perl6-dev
11:16 AlexDaniel joined #perl6-dev
11:34 squashable6 joined #perl6-dev
11:55 Ven joined #perl6-dev
12:13 lizmat https://blog.toggl.com/world-created-programmer/  # made me lol
12:15 Zoffix This rings too true to be funny :) "To create darkness, you need to hide the land, but that's OK cause creatures won't be added until later"
12:37 llfourn joined #perl6-dev
12:51 moritz lizmat: me too. Thanks!
13:05 Geth ¦ nqp/master: 5 commits pushed by pmurias++
13:05 Geth ¦ nqp/master: 66caf96673 | [js] Fix mixing default positional with slurpy positional
13:05 Geth ¦ nqp/master: 7f05f60b5c | Test mixing slurpy positionals with default positionals
13:05 Geth ¦ nqp/master: fcce8ab5ce | [js] Multi cache native refs separately
13:05 Geth ¦ nqp/master: 4e58cc54a5 | [js] Add types to variables
13:05 Geth ¦ nqp/master: 87000502c5 | [js] Stop using a dynamic variable when a regular one is enough
13:05 Geth ¦ nqp/master: review: https://github.com/perl6/nqp/compare/5f9e354ccf...87000502c5
13:26 MasterDuke anyone have an opinion on https://github.com/rakudo/rakudo/pull/1221/? if it is going to be merged, it'll be easier to do it now rather than later, since it touches so many lines it's fragile to conflicts from other changes
13:29 MasterDuke pmurias: i added the new nqp::fromI_I recently for moar and jvm, but i don't know javascript enough to do it for that backend
13:30 perlpilot MasterDuke: my opinion:  +1   :)
13:34 MasterDuke cool
13:38 Zoffix MasterDuke: is that just for consistency?
13:38 Zoffix Or will `{ * }` stop working soon and this is to avoid that breakage? I thougth I saw something along those lines mentioned somewhere
13:39 MasterDuke `{ * }` breaks if you add scope around it
13:39 Zoffix Ah
13:39 MasterDuke i don't think it's intended to break otherwise
13:41 Geth ¦ rakudo: 0edd0cc9c1 | MasterDuke17++ | 23 files
13:41 Geth ¦ rakudo: Normalize proto bodies
13:41 Geth ¦ rakudo:
13:41 Geth ¦ rakudo: Convert all `{ * }` to `{*}` for consistency.
13:41 Geth ¦ rakudo: review: https://github.com/rakudo/rakudo/commit/0edd0cc9c1
13:41 Geth ¦ rakudo: 733d50c03d | MasterDuke17++ (committed using GitHub Web editor) | 23 files
13:41 Geth ¦ rakudo: Merge branch 'master' into normalize_proto_bodies
13:41 Geth ¦ rakudo: review: https://github.com/rakudo/rakudo/commit/733d50c03d
13:41 Geth ¦ rakudo: 2cadb2bc01 | (Zoffix Znet)++ (committed using GitHub Web editor) | 23 files
13:41 Geth ¦ rakudo: Merge pull request #1221 from MasterDuke17/normalize_proto_bodies
13:41 synopsebot RAKUDO#1221 [closed]: https://github.com/rakudo/rakudo/pull/1221 Normalize proto bodies
13:41 Geth ¦ rakudo:
13:41 Geth ¦ rakudo: Normalize proto bodies
13:41 Geth ¦ rakudo: review: https://github.com/rakudo/rakudo/commit/2cadb2bc01
13:42 MasterDuke thanks
13:43 perlpilot MasterDuke: I even think that `proto sub foo(|) {*}` would be better written as `proto sub foo(|) { {*} }`  but that might just be some foolish consistency on my part :)
13:43 perlpilot (I think the latter helps remind that there can be other code in the proto)
13:44 perlpilot and {*} makes a nice visual pill
13:46 MasterDuke perlpilot: that does have its appeal...but i wonder if the extra scope would noticeably slow down parsing and/or execution?
13:47 Zoffix A quick glance at optimized AST tells me `{ {*} }` might not be considered an "only star" proto and misses some optimizations
13:48 perlpilot Then, yeah, it's a foolish consistency on my part :)
13:49 Zoffix Don't see any dif in time measurements tho
13:49 Zoffix (or maybe it's just drowned by noise of the bench
13:49 Zoffix )
13:50 Zoffix Ah nah
13:50 Zoffix m: proto z(|) { * }; multi z ($) { }; for ^1000_000 { $ = z 42;  }; say now - INIT now
13:50 camelia rakudo-moar 264eef798: OUTPUT: «0.1827956␤»
13:50 Zoffix m: proto z(|) { {*} }; multi z ($) { }; for ^1000_000 { $ = z 42;  }; say now - INIT now
13:50 camelia rakudo-moar 264eef798: OUTPUT: «0.5886782␤»
13:50 Zoffix m: say .58867/.1827
13:50 camelia rakudo-moar 264eef798: OUTPUT: «3.222058␤»
13:52 MasterDuke wow, didn't expect it'd be quite that big
13:55 * jnthn isn't surprised
13:56 jnthn The first form can skip the call to the proto entirely once it's cached
13:56 jnthn The second form can't
14:00 Geth ¦ rakudo: fccc75152d | (Elizabeth Mattijsen)++ | lib/Telemetry.pm6
14:00 Geth ¦ rakudo: Quite extensive refactor of Telemetry
14:00 Geth ¦ rakudo:
14:00 Geth ¦ rakudo: - just store the nqp::getrusage struct instead of already dissecting
14:00 Geth ¦ rakudo:   - makes Telemetry.new about 15% faster
14:00 Geth ¦ rakudo:   - adapt all instance method logic accordingly
14:00 Geth ¦ rakudo:   - moves burden to Telemetry::Period creation, but that's when we need
14:00 Geth ¦ rakudo:     and all of the snapping is already done
14:00 Geth ¦ rakudo: <…commit message has 10 more lines…>
14:00 Geth ¦ rakudo: review: https://github.com/rakudo/rakudo/commit/fccc75152d
14:28 dogbert17 ZOFFLOP: t/spec/S11-modules/nested.t
14:29 dogbert17 there's something fishy with this test, it failed under spectest and when I ran it standalone I got (on the first run): Use of uninitialized value $repo-id of type Any in string context.
14:29 dogbert17 Methods .^name, .perl, .gist, or .say can be used to stringify it to something meaningful.
14:29 dogbert17 in any statement_control at blib/Perl6/Grammar.moarvm line 1
14:33 Zoffix I observed the same behaviour for it during my flops
14:42 Ven joined #perl6-dev
14:43 dogbert17 Zoffix: are you still experiencing RT #132273 ?
14:43 synopsebot RT#132273 [new]: https://rt.perl.org/Ticket/Display.html?id=132273 [SEGV] while running t/spec/S17-supply/supplier-preserving.t
14:45 Zoffix haven't tried
14:45 dogbert17 ok, just checking :)
14:45 Zoffix |6h https://rt.perl.org/Ticket/Display.html?id=132273
14:45 ZofBot Zoffix, Will remind you on 2017-11-03T16:45:22.427746-04:00 about https://rt.perl.org/Ticket/Display.html?id=132273
14:46 dogbert17 will try to repro but suspect that I'll fail
14:52 ggoebel joined #perl6-dev
14:58 lizmat m: constant method = 1; my @a; @a[method] = 42   # DIHWIDT  :-(
14:58 camelia rakudo-moar fccc75152: OUTPUT: «5===SORRY!5=== Error while compiling <tmp>␤Missing block␤at <tmp>:1␤------> 3constant method = 1; my @a; @a[method7⏏5] = 42   # DIHWIDT  :-(␤»
15:01 Zoffix m: constant method = 1; my @a; @a[::method] = 42; say @a
15:01 camelia rakudo-moar fccc75152: OUTPUT: «[(Any) 42]␤»
15:01 Zoffix m: constant constant = 1; constant @a = constant; say @a
15:01 camelia rakudo-moar fccc75152: OUTPUT: «(1)␤»
15:02 Zoffix ^_^
15:14 AlexDaniel joined #perl6-dev
15:43 ugexe what is List.to / List.from intended to solve?
15:44 [Coke] |1m /me wonders what this does.
15:45 ugexe it expects either the first or last item of a list to be a specific type (Match), and one works on the first item and one works on the last
15:46 Geth ¦ rakudo: 1c2c7d845e | (Elizabeth Mattijsen)++ | lib/Telemetry.pm6
15:46 Geth ¦ rakudo: Added formatting capabilities to Telemetry report
15:46 Geth ¦ rakudo:
15:46 Geth ¦ rakudo: - :@format named parameter
15:46 Geth ¦ rakudo: - expects a list with lists of: column name, method name, format, legend
15:46 Geth ¦ rakudo:   - format *without* the preceding '%'
15:46 Geth ¦ rakudo: review: https://github.com/rakudo/rakudo/commit/1c2c7d845e
15:46 * [Coke] has been spending way too much time at $dayjob lately, and misses hacking on p6. keep up the good work, everyone. ;)
15:46 lizmat gfldex: ^^^
15:46 lizmat afk for a bit&
15:46 ugexe m: my @a = ("a" ~~ /a/, "bb" ~~ /b/); say @a.from; say @a.to
15:46 camelia rakudo-moar fccc75152: OUTPUT: «0␤1␤»
15:46 ugexe m: my @a = ("a" ~~ /a/, "bb" ~~ /a/); say @a.from; say @a.to
15:46 camelia rakudo-moar fccc75152: OUTPUT: «0␤No such method 'to' for invocant of type 'Any'. Did you mean 'so'?␤  in block <unit> at <tmp> line 1␤␤»
15:47 ugexe that seems to indicate its not very robust
15:48 jnthn ugexe: m:g
15:48 ugexe right, but even the error message implies something strange
15:48 ugexe its calling .to on a List, but the error complains about calling .to on an item in the list items
15:49 jnthn m:g will only ever produce a list of Match objects
15:49 evalable6 jnthn, rakudo-moar fccc75152: OUTPUT: «(exit code 1) ===SORRY!=== Error while compiling /tmp/kGiCf5ENky␤Missing…»
15:49 evalable6 jnthn, Full output: https://gist.github.com/716eef5e734b1efc50252d1fe93826f6
15:50 Zoffix ugexe: IIRC it's 'cause List.to/.from are an, awful hackl
15:50 Zoffix s: (), 'to', \()
15:50 SourceBaby Zoffix, Sauce is at https://github.com/rakudo/rakudo/blob/4fca94743/src/core/List.pm#L372
15:50 Zoffix It basically *assumes* if such a method is called, the list is a list of Match objects
15:50 ugexe thats why i wonder what its useful for
15:50 * AlexDaniel slaps evalable6
15:50 ugexe i assume its useful for -something-
15:51 * Zoffix has no idea what for and wouldn't mind the hack going away :P
15:52 Zoffix But I'm guessing it's for uniform use with $/.to and $/.from giving the right result regardless of whether the $/ has a list of Matches (e.g. from + quantifier) or just one Match
15:52 ugexe then maybe its not needed since the cursor rework stuff awhile back
15:54 Zoffix m: 'abc' ~~ /(.)+/; say $/[0].^name
15:54 camelia rakudo-moar 1c2c7d845: OUTPUT: «Array␤»
15:54 Zoffix Wouldn't it need to be here ^
15:54 Zoffix m: 'abc' ~~ /(.)+/; say [.from, .to] with $/[0]
15:54 camelia rakudo-moar 1c2c7d845: OUTPUT: «[0 3]␤»
16:22 ugexe removing it causes two spectest files to fail, but it still passed the Grammar::HTTP tests
16:24 Ven joined #perl6-dev
16:30 mst if unicode gets more than one pile of poo glyph can we have an [[:excremoji:]] matcher for it?
16:30 Zoffix :}
16:41 ilmari \p{oo}
16:41 ilmari \poo even
16:46 buggable New CPAN upload: cro-0.7.1.tar.gz by JNTHN https://cpan.metacpan.org/authors/id/J/JN/JNTHN/Perl6/cro-0.7.1.tar.gz
16:46 buggable New CPAN upload: cro-core-0.7.1.tar.gz by JNTHN https://cpan.metacpan.org/authors/id/J/JN/JNTHN/Perl6/cro-core-0.7.1.tar.gz
16:46 buggable New CPAN upload: cro-http-0.7.1.tar.gz by JNTHN https://cpan.metacpan.org/authors/id/J/JN/JNTHN/Perl6/cro-http-0.7.1.tar.gz
16:46 buggable New CPAN upload: cro-tls-0.7.1.tar.gz by JNTHN https://cpan.metacpan.org/authors/id/J/JN/JNTHN/Perl6/cro-tls-0.7.1.tar.gz
16:46 buggable New CPAN upload: cro-websocket-0.7.1.tar.gz by JNTHN https://cpan.metacpan.org/authors/id/J/JN/JNTHN/Perl6/cro-websocket-0.7.1.tar.gz
16:46 buggable New CPAN upload: cro-zeromq-0.7.1.tar.gz by JNTHN https://cpan.metacpan.org/authors/id/J/JN/JNTHN/Perl6/cro-zeromq-0.7.1.tar.gz
16:46 timotimo nice
16:48 * Zoffix ponders what could be put into cro-magnon namespace...
16:48 Zoffix https://en.wikipedia.org/wiki/Cro-Magnon
16:55 Zoffix ZofBot: the humans did not appreciate the joke about their ancenstor. Abort Mission!
16:55 ZofBot Zoffix, When the time came for me to join the firm, I put it to dad straight
16:56 Zoffix Didn't know you had a dad, bruh.
16:58 [Coke] is it intentional to get cpan upload info to both #perl6 & #perl6-dev ?
17:04 ilmari Zoffix: I thought you were the dad?
17:04 Zoffix [Coke]: yes. My logic is not everyone is present in both of those channels to select just a single one to have those reports appear in.
17:04 Zoffix ilmari: it's complicated! :o
17:04 * ilmari likes how Jeph Jacques refers to his ebooks bot as "my idiot robot son"
17:05 Zoffix heh
17:05 ilmari https://twitter.com/jephjacques / https://twitter.com/jeph_ebooks
17:05 Zoffix [Coke]: you're the second person to ask that tho. I think I'll move #perl6-dev reports to #zofbot...
17:06 [Coke] eh. irc sucks. I get it. :)
17:07 buggable joined #perl6-dev
17:07 Zoffix Done.
17:21 El_Che any alpine devs around?
17:32 robertle joined #perl6-dev
17:41 greppable6 joined #perl6-dev
18:00 AlexDaniel joined #perl6-dev
18:05 Geth ¦ rakudo: MasterDuke17++ created pull request #1229: Test for RT #132291
18:05 Geth ¦ rakudo: review: https://github.com/rakudo/rakudo/pull/1229
18:05 synopsebot RT#132291 [new]: https://rt.perl.org/Ticket/Display.html?id=132291 [LTA] Too few positionals passed to '' (for [:a] X [:b] -> ($i, $j) { })
18:05 synopsebot RAKUDO#1229 [open]: https://github.com/rakudo/rakudo/pull/1229 Test for RT #132291
18:07 Geth ¦ rakudo: cd9f66eff5 | MasterDuke17++ | t/05-messages/02-errors.t
18:07 Geth ¦ rakudo: Test for RT #132291
18:07 Geth ¦ rakudo: review: https://github.com/rakudo/rakudo/commit/cd9f66eff5
18:07 Geth ¦ rakudo: 32412170b3 | (Zoffix Znet)++ (committed using GitHub Web editor) | t/05-messages/02-errors.t
18:07 Geth ¦ rakudo: Merge pull request #1229 from MasterDuke17/test_for_RT132291
18:07 Geth ¦ rakudo:
18:07 Geth ¦ rakudo: Test for RT #132291
18:07 Geth ¦ rakudo: review: https://github.com/rakudo/rakudo/commit/32412170b3
18:11 El_Che I am working on alpine packages, and I have a failing test: https://gist.github.com/nxadm/f1c10be2a03b22c34471345fbde35c72 ( t/04-nativecall/01-argless.t)
18:11 El_Che I know alpine is not the nicest for C and C++ code, but I haven't seen this one
18:11 El_Che Failed test 'called argless function returning char' at t/04-nativecall/01-argless.t line 25 expected: '2' got: '94117762201346'
18:12 El_Che if someone has seen it, ping me
18:12 Zoffix m: say 94117762201346.log: 2
18:12 camelia rakudo-moar 32412170b: OUTPUT: «46.4195322521493␤»
18:12 * Zoffix shrugs
18:12 El_Che good ... how :)
18:13 AlexDaniel MasterDuke: :D
18:14 AlexDaniel ( https://rt.perl.org/Ticket/Display.html?id=132291#txn-1504393 )
18:15 AlexDaniel I hope twice is enough! :)
18:15 MasterDuke heh, that's one thoroughly resolved ticket
18:16 El_Che is there a 2017.10.1 coming or is it for 2017.11?
18:16 El_Che I asked becuase fpm (what I use to generate pkgs) is broken on alpine
18:17 El_Che so I went into the rabbit hole to create the pkg manually
18:17 El_Che using the tools (that downloads and compiles the stuff for you) you can get a package, but you loose a lot of flexibility
18:17 El_Che I don't want to figure out how patches work on the alpine build system :)
18:18 MasterDuke i don't think there are any plans for a 2017.10.1
18:18 El_Che there are no alpine packages, so it's not like it's blocking or something :)
18:18 Zoffix 2017.11 is gonna be here in just 2 weeks
18:18 El_Che indeed
18:34 AlexDaniel El_Che: what's the ticket id for that issue?
18:35 AlexDaniel releasable6: next
18:35 releasable6 AlexDaniel, Next release in 15 days and ≈0 hours. No blockers. 0 out of 113 commits logged
18:35 releasable6 AlexDaniel, Details: https://gist.github.com/03c27e76a8a89b9c11fe7f49b6de1969
18:37 El_Che AlexDaniel: 13229
18:37 AlexDaniel RT#13229
18:37 AlexDaniel mm
18:37 AlexDaniel ?
18:38 Zoffix There ain't no such ticket. The number's missing a digit
18:39 El_Che 132291
18:39 El_Che sorry
18:39 bisectable6 joined #perl6-dev
18:39 Zoffix RT#132291
18:39 synopsebot RT#132291 [resolved]: https://rt.perl.org/Ticket/Display.html?id=132291 [LTA] Too few positionals passed to '' (for [:a] X [:b] -> ($i, $j) { })
18:40 Zoffix El_Che: are you sure? That's just some LTA error, why's that blocking alpine?
18:40 Geth ¦ nqp/master: 4 commits pushed by pmurias++
18:40 Geth ¦ nqp/master: be6a7a167d | [js] Implement nqp::fromI_I
18:40 Geth ¦ nqp/master: 1639c2b87a | Test nqp::fromI_I
18:40 Geth ¦ nqp/master: c3e3d9bb25 | [js] Check that no unhandled named arguments are passed
18:40 Geth ¦ nqp/master: 20536d98a5 | Test that unhandled named arguments to subs are not allowed
18:40 Geth ¦ nqp/master: review: https://github.com/perl6/nqp/compare/87000502c5...20536d98a5
18:40 El_Che Zoffix: i meant the other way round
18:41 El_Che Zoffix: as in "it's literally not important"
18:41 El_Che I am working on a first alpine pkg
18:41 El_Che so it's ok if the fix in in 2017.11 or later
18:41 * Zoffix is confused
18:41 Zoffix El_Che: fix for what?
18:42 Zoffix For alpine issue? Then the question is: what's that issue? What's the ticket ID for it?
18:42 El_Che ok, let me rephrase, read as no-sarcasm
18:42 El_Che so, there is a bug that make a test fail, that is already fixed in rakudo upstream
18:43 El_Che I didn't know that
18:43 Zoffix Ah ok
18:43 El_Che So, now I'll continue without make test
18:43 El_Che in order to figure the alpine packaging out
18:43 El_Che by the time I have a pkg, rakudo 2017.11 will be out and the tests will pass
18:44 AlexDaniel El_Che: OK cool. If you notice any further issues on alpine, please file a bug report and ping me the ticket ID. I'll make sure we do something about it before the release.
18:44 El_Che I am half way in understanding how alpine building works :)
18:45 evalable6 joined #perl6-dev
19:12 travis-ci joined #perl6-dev
19:12 travis-ci NQP build failed. pmurias 'Test that unhandled named arguments to subs are not allowed'
19:12 travis-ci https://travis-ci.org/perl6/nqp/builds/296926994 https://github.com/perl6/nqp/compare/87000502c5ef...20536d98a51e
19:12 travis-ci left #perl6-dev
19:48 El_Che ok, --gen-* is a no-no on alpine as it tries to write in the --prefix dir before make install
20:03 releasable6 joined #perl6-dev
20:12 lx joined #perl6-dev
20:15 Geth ¦ rakudo: 8b24bf5cd4 | usev6++ | src/vm/jvm/runtime/org/perl6/rakudo/RakudoContainerSpec.java
20:15 Geth ¦ rakudo: [jvm] Also do a type check when assigning Nil
20:15 Geth ¦ rakudo:
20:15 Geth ¦ rakudo: Fixes GH#1225, pmurias++
20:15 Geth ¦ rakudo: review: https://github.com/rakudo/rakudo/commit/8b24bf5cd4
20:15 synopsebot GH#1225 [open]: https://github.com/rakudo/rakudo/issues/1225 [JVM] Do a type check when resetting a default when asigning a Nil
20:57 b2gills left #perl6-dev
21:06 b2gills joined #perl6-dev
21:17 Zoffix dogbert17: RT#132273 no longer segfaults; closed ticket
21:17 synopsebot RT#132273 [resolved]: https://rt.perl.org/Ticket/Display.html?id=132273 [SEGV] while running t/spec/S17-supply/supplier-preserving.t
21:50 jnthn evening o/
21:53 Zoffix \o
22:06 jnthn https://github.com/rakudo/rakudo/issues/1230 is the bug I ran into the other day
22:06 jnthn Filed it so we don't forget on it, but intend to work on it next week or so
22:08 jnthn Did we pick a way to refer to GitHub issues from roast?
22:08 * Zoffix just used full URL so far
22:09 Zoffix Also makes GitHub include reference to the commit on the issue
22:10 Zoffix Like here: https://github.com/perl6/roast/blob/master/S04-statements/return.t#L106
22:11 Zoffix Oh the ref happens if you also include link in commit
22:11 jnthn Alright, will follow that convention
22:11 Zoffix or write "rakudo/1234"
22:11 Zoffix (in commit message I meant)
22:12 jnthn I was mostly thinking about the test file itself
22:12 jnthn But yeah, "Test for rakudo/1234" getting linked in the commit message is handy
22:14 jnthn hah, phew...the heap of Travis fail notifications I got for Cro were because there was some minutes between me removing META.list entries and the CPAN additions being visible to zef
22:20 El_Che About travis. I now also create Ubuntu 14.04 rakudo debs (OS that travis run), so I can speedup tests of my stuff: https://github.com/nxadm/rakudo-pkg/blob/master/.travis.yml#L27 . Is there a use case for me providing debs for older rakudo releases? Maybe in a repo? Or is there no use case before rakudbrew is more flexible?
22:20 El_Che s/before/because/
22:29 jnthn oh heck, I must update some of mine for that
22:29 jnthn El_Che: Do you do 16.04 ones, ooc?
22:29 jnthn Oh wait, Travis doesn't support that yet
22:29 jnthn Heh
22:30 jnthn El_Che: But in terms of "older releases" - yes, for example I'd like to be able to test against the last Rakudo that was used to make an R*.
22:31 El_Che well, 14.04 shoould die in 2019 :)
22:31 jnthn I don't think Travis does 16.04 yet, though.
22:31 El_Che no, travis is 14.04, hence the new 14.04 pkgs
22:33 jnthn Once there's a 6.d language release, most probably the first release of Rakudo that supported that will be an interesting thing to test against long-term too
22:35 El_Che well, from 2017.10 onwards there will be a deb for 14.04
22:35 El_Che it's automated :)
22:35 jnthn Ah, I see
22:36 El_Che my point is that regular modules probably wont test against random commits
22:36 jnthn For me personally, that'll suffice
22:36 El_Che so If we can provide a matrix of debs, it will shave 10minutes of every build
22:38 El_Che jnthn: cheers on cro, btw. I just had a look at the docs
22:38 El_Che fancy site :)
22:50 jnthn :)
22:50 jnthn Site built using Cro, of course :)
22:52 El_Che I remember the crust author talking about threading problems at fosdem
22:52 El_Che you skip that middellayer?
22:53 El_Che it was this talk: https://archive.fosdem.org/2016/schedule/event/crust/
22:53 jnthn Well, between then and now there's been a lot of fixes also
22:54 jnthn Cro uses IO::Socket::Async though
22:54 jnthn It doesn't do any explicit threading; it's built entirely out of the higher-level Perl 6 constructs
22:55 jnthn Meaning that a server under low load will run with just a handful of threads, and one under load will happily start some more
22:56 Geth ¦ roast: fd3ba22c63 | usev6++ | S32-io/io-handle.t
22:56 Geth ¦ roast: [jvm] Fudge failing tests in S32-io/io-handle.t
22:56 Geth ¦ roast: review: https://github.com/perl6/roast/commit/fd3ba22c63
22:56 jnthn Cro so far takes the Node.js approach of "run our own HTTP server"
22:57 El_Che jnthn: or go's
22:57 jnthn That too
22:57 jnthn Or .Net or... :)
22:57 jnthn That said, Cro is just a way to build async pipelines
22:57 gfldex i'm afraid of using Perl 6 in a coding competion in fear of being accused of cheating. :->
22:58 jnthn So there's no reason at all that we couldn't provide FastCGI support
22:58 El_Che jnthn: I heard an interesting talk about the jvm and docker at oss summit last week. About how only java 9 was party aware of limit set to the container
22:58 El_Che and I wondered how rakudo worked
22:59 El_Che e.g. problem was that the jvm saw x cores, while it only had access to let's say 2
22:59 El_Che but because it saw all the cores in the machines it started X threads per core for housekeeping
22:59 jnthn We're not quite that aggressive.
22:59 El_Che so each container used way to much memory
23:00 jnthn But still somewhat vulnerable to that issue
23:00 El_Che java 9 now can take the heap hint from the container limits, and support dedicated cores (not percentages, what most people use)
23:00 jnthn MoarVM only starts one or two internal threads
23:01 El_Che jnthn: that's ok
23:01 jnthn So that's OK
23:01 jnthn Rakudo's thread pool scheduler is more interesting.
23:01 El_Che indeed
23:01 jnthn It will add threads if it sees that there's more work than the current are rapidly chewing through
23:01 jnthn Up until the CPU core count
23:01 jnthn It probably, though we'd have to check this, seeing the same number the JVM was
23:02 El_Che so it can exhaust the ram
23:02 jnthn Beyond the CPU core count it'll reluctantly add threads if it things the program is deadlocked
23:02 El_Che resulting in OOM for the container
23:02 jnthn *nod*
23:03 jnthn It's probably worth factoring memory into the scheduler's decision making too
23:03 El_Che I think that probably 70% of the talks at linux con were about containers
23:03 El_Che so that's something to keep in mind
23:04 jnthn For sure
23:04 El_Che besides that containers make the more fragile dependency installation less of an issue
23:04 El_Che kind-of levels the playfiel
23:04 El_Che d
23:05 jnthn Yeah; we'll be providing Cro docker images in the near future :)
23:05 El_Che I want to provide rakudo alpine pkgs
23:05 El_Che so paranoid people like me can build their own cro images
23:05 jnthn :)
23:06 jnthn I'll probably look for a base image that has an OpenSSL with ALPN support, so folks can get HTTP/2.0 support out of the box with them
23:06 El_Che centos 7.4?
23:07 El_Che centos, debian, ubuntu are popular and sane picks
23:07 El_Che but alpine is sooo small :)
23:07 jnthn I'm most likely to go debian or ubuntu, out of familiarity more than anything :)
23:07 El_Che I run production on centos 7 images though
23:07 El_Che linux team run centos
23:07 El_Che otherwise would have been debian/ubuntu
23:08 El_Che I have a few things in alpine though
23:08 El_Che small apps in prod
23:08 El_Che specially good for go code
23:08 El_Che I wonder how alpine will be for the perl 6 ecosystem
23:08 El_Che some things are tough to compile
23:09 jnthn Yeah, we had to patch MoarVM a bit because of its small stack size
23:09 El_Che yes, I remember that
23:10 jnthn With regards to the thread scheduler, I have the unenviable task of trying to make it sufficiently smart and also sufficiently understandable.
23:10 jnthn Because one can never make it so smart that nobody will ever thing to ask how it makes its choices :)
23:10 jnthn *think
23:10 El_Che bus bus bus
23:10 El_Che :)
23:11 jnthn And it used to be so stupid I could explain, "well, it just predictably does this really awful thing" :)
23:11 jnthn And now it's more like "It tries to do X, and it depends on Y and Z" :)
23:11 El_Che soon you'll need those 3D screens from science fiction movies
23:11 jnthn There are at least env vars one can set to see the data and decisions though
23:12 El_Che "I see dead threads"
23:38 Zoffix So a wild guess: edits to issues aren't re-sent to people who only read them via email, right?
23:38 Zoffix On GitHub
23:39 jnthn Zoffix: Matches what I've observed, yes

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