timotimo github.com/XiKuuKy/Avolution-Emoji.../Emoji.pm6 - petition to add a :ignorecase and perhaps :ignoremark parameters to method trans 08:53
which will likely cause :i and :m to also appear in Str.match 08:59
[Tux] Rakudo version 2020.01-91-g99b67a4b6 - MoarVM version 2020.01.1-29-g0dd136440
csv-ip5xs0.730 - 0.802
csv-ip5xs-206.193 - 6.315
csv-parser22.303 - 22.612
csv-test-xs-200.367 - 0.390
test7.661 - 7.687
test-t1.807 - 2.138
test-t --race0.906 - 0.908
test-t-2031.765 - 32.197
test-t-20 --race9.105 - 9.128
10:13
AlexDaniel so gist.github.com/a60322ff73b1faac22...b6c0478b68 and gist.github.com/484c4b653cbd936240...5cdaf78723 12:17
lizmat: maybe we should, uh… stop squashathons altogether? Some months they were great, but lately nobody cares that much
Geth z: stoned++ created pull request #7:
Update doc repository URL
12:41
AlexDaniel hm, we should move nqp and roast 12:43
lizmat hmmm... t/spec/S17-supply/syntax.t hung last night on make spectest 14:38
AlexDani` agree that maybe we should doing them until there are enough people wanting to do them 14:39
also: personally, weekends don't work too well for me
well, haven't in the past months 14:40
and maybe we should expand it to a whole week, at a lower level
to give more people a chance to work on it
emphasising the "marathon" part rather than the "squash" part 14:41
MasterDuke a while ago i suggested slowing down to something like quarterly. two months break, 3 weeks of advertisement/prep, 1 weekend (or maybe the whole week) of activity 14:43
i had a segfault a day or two ago when building nqp. tried running the step that failed in a loop for a while with spectests going on in the background, but couldn't repro 14:44
Geth rakudo: 497afdaded | (Elizabeth Mattijsen)++ | src/core.c/Date.pm6
Make Date.new(Str) about 3.5x as fast still

  - fast-path well-formed yyyy-mm-dd without using a regex
  - fallback to regex if not well formed
  - use Str.match(/.../) rather than ~~ /.../
  - profit!
15:07
lizmat that makes it about 10x as fast from before this weekend 15:08
nine MasterDuke: I've seen segfaults in NQP and rakudo on the OBS, too. But of course none are reproducible
AlexDaniel MasterDuke: well, it's already once every two months 15:14
but yeah
MasterDuke nine: sounds like some of memory corruption or race condition in moarvm. i've also seen some flapping in the precomp tests, which seems like it might be related 15:34
travis-ci Rakudo build failed. Elizabeth Mattijsen 'Make Date.new(Str) about 3.5x as fast still 16:09
travis-ci.org/rakudo/rakudo/builds/648045611 github.com/rakudo/rakudo/compare/9...7afdadedbb
lizmat restarted failing job 16:10
Geth rakudo: 5b97ddefe7 | (Elizabeth Mattijsen)++ | 2 files
Introduce R:I.posix-and-leap-from-tai

  - formerly known as posix-from-tai
  - repurpose posix-from-tai to just returning epoch without leap-second info
  - adapt caller as appropriate
16:34
MasterDuke lizmat: "take TAI, return epoch and if in leap-second" typo? 16:38
lizmat and "if in leap second" clearer ? 16:39
MasterDuke think so 16:40
Geth rakudo: bd948ba31d | (Elizabeth Mattijsen)++ | src/core.c/Instant.pm6
Eliminate Instant.SET-SELF

  - replace by nqp::p6bindattrinvres
  - saves a few allocations for each Instant creation
  - makes all Instants and Instant operations marginally faster
16:53
Geth nqp: 6ab38d43b4 | (Elizabeth Mattijsen)++ | tools/templates/MOAR_REVISION
Bump MoarVM to get libtommath leak fixes

  MasterDuke++
17:07
MasterDuke it would be nice if sena_kun++ could do a Blin run soon. there have been some big changes recently 17:11
sena_kun MasterDuke: I did last night, just not reviewed the results yet 17:14
MasterDuke great timing
sena_kun :)
Geth rakudo: 9a83b61e46 | (Elizabeth Mattijsen)++ | tools/templates/NQP_REVISION
Bump NQP to get latest libtommath fixes

  MasterDuke++
17:20
Geth rakudo: e2a66ff61c | (Daniel Green)++ | t/09-moar/UnipropCheck.pm6
Correct the todo count for some Unicode tests
17:35
rakudo: e979f17338 | (Elizabeth Mattijsen)++ | src/core.c/Date.pm6
Abstract Date.new-from-daycount logic into private method

  - typical case of copy-paste without much reason for it
17:43
Geth rakudo: af8cca773e | (Elizabeth Mattijsen)++ | src/core.c/Date.pm6
Date!new-from-daycount no longer uses !SET-SELF

  - since we know that all values are valid, no checks needed
  - allows removal of the optional $daycount parameter of SET-SELF
  - makes all Date objects marginally faster (one less onbject/one less check)
17:57
Geth rakudo: 65d0422b2b | (Elizabeth Mattijsen)++ | src/core.c/Rakudo/Internals.pm6
Add R:I.epoch-from-tai

Basically the same as posix-from-tai, but only the integer part, which allows for simpler calculation.
18:49
lizmat hmmm.. just had a hang on t/spec/S32-io/IO-Socket-Async.t, test 36: both receivers finished without exception 18:53
Geth rakudo: f91a16169a | (Elizabeth Mattijsen)++ | src/core.c/Date.pm6
Make Date.new(Instant) 12.5x as fast

  - by skipping the DateTime intermediate object creation
  - using the new R:I.epoch-from-tai function
18:56
Geth rakudo: 1d7a91689a | (Elizabeth Mattijsen)++ | src/core.c/Date.pm6
Make Date.today about 3x as fast

  - by skipping the DateTime intermediate object creation
  - using the new R:I.epoch-from-tai function
19:26
Geth rakudo: 61821ffa17 | (Elizabeth Mattijsen)++ | 2 files
Change CompileTestLib extension to .rakumod
19:36
rakudo: 8c242d5e7b | (Elizabeth Mattijsen)++ | 2 files
Change UnipropCheck extension to .rakumod
19:37
rakudo: 582c14a81c | (Elizabeth Mattijsen)++ | 2 files
Change CustomOps extension to .rakumod
19:38
Geth rakudo: 4ebbddfbf8 | (Elizabeth Mattijsen)++ | src/core.c/Date.pm6
Make Date.later/earlier between 1.5x and 5.5x as fast

  - depending on the unit chosen
  - make "earlier" a first class citizen
  - abstract actual work into several private methods
21:53
sena_kun ok, blin results are sorted out: no huge issues, the "bisected" are around the normal cases, submitted a new issue 22:08
lizmat sena_kun: on a module, or on rakudo ? 22:10
sena_kun lizmat: github.com/rakudo/rakudo/issues/3468 is for rakudo 22:12
sena_kun sees a lot of cool PRs 22:18
Geth rakudo: 5943cbfe02 | (Elizabeth Mattijsen)++ | src/core.c/Date.pm6
Make sure Date.earlier/later work with subclasses

Instead of creating a new object, clone the existing object and update the appropriate attributes.
22:21
sena_kun I have no rights to merge things like github.com/rakudo/rakudo/pull/3361/ ? inb4: of course no
Geth rakudo: 91f8943597 | (Bahtiar Gadimov)++ (committed by Bahtiar `kalkin-` Gadimov) | lib/Pod/To/Text.rakumod
Pod::To::Text display subset documentation
22:23
rakudo: 94b30b667b | (Elizabeth Mattijsen)++ (committed using GitHub Web editor) | lib/Pod/To/Text.rakumod
Merge pull request #3361 from kalkin/Pod-To-Text-Subsets

Pod::To::Text display Subsets
sena_kun github.com/rakudo/rakudo/pull/3360 is the same level and github.com/rakudo/rakudo/pull/3369 (this one wants a bit more attention, adds a new sub) 22:24
lizmat++
Geth rakudo: f60d969443 | (Bahtiar `kalkin-` Gadimov)++ | lib/Pod/To/Text.rakumod
Pod::To::Text display enums
rakudo: ca912be1bf | (Elizabeth Mattijsen)++ (committed using GitHub Web editor) | lib/Pod/To/Text.rakumod
Merge pull request #3360 from kalkin/Pod-To-Text-Enums

Pod::To::Text display enums
rakudo: 0b0a1aad50 | (Bahtiar Gadimov)++ (committed by Bahtiar `kalkin-` Gadimov) | lib/Pod/To/Text.rakumod
Fix #2371 - Pod::To::Text display sub & method return type
22:26
rakudo: a152997d13 | (Elizabeth Mattijsen)++ (committed using GitHub Web editor) | lib/Pod/To/Text.rakumod
Merge pull request #3369 from kalkin/sub-and-method-return-type

Fix #2371 - Pod::To::Text display sub & method return type
linkable6 RAKUDO#2371 [closed]: github.com/rakudo/rakudo/issues/2371 [POD6] Pod declarator block does not show sub return type
sena_kun lizmat is awesome. \o/ 22:27
Geth rakudo: 9707384ce5 | (Elizabeth Mattijsen)++ | src/core.c/Date.pm6
Make Date.succ/pred about 2.8x as fast

  - add helper methods for moving days
  - let nqp::clone take care of subclasses
23:01
japhb lizmat++ # Much faster Date manipulation 23:17
lizmat bisectable6: dd Date.new(Instant.from-posix(<3300154654195/2087>, Bool::False)) 23:33
bisectable6 lizmat, Bisecting by exit code (old=2015.12 new=9707384). Old exit code: 1
lizmat, bisect log: gist.github.com/2ae9008256256f59e0...99598622e3
lizmat, (2016-02-07) github.com/rakudo/rakudo/commit/27...5e7412c95a
lizmat bisectable6: old=2020.01 dd Date.new(Instant.from-posix(<3300154654195/2087>, Bool::False)) 23:34
bisectable6 lizmat, On both starting points (old=2020.01 new=9707384) the exit code is 0 and the output is identical as well
lizmat, Output on both points: «Date.new(2020,2,9)␤»
lizmat hmmm... looks like we have a spectest that fails (for me) between midnight and 1am
Geth rakudo: cfe2cdc567 | (Elizabeth Mattijsen)++ | src/core.c/Date.pm6
Make Date integer arithmetic about 2.3x as fast
23:35
lizmat and that concludes my hacking for today& 23:36