01:24 MasterDuke left 01:33 lucasb left 03:37 leont left
Geth rakudo: vrurg++ created pull request #4082:
Fix handling of definite parameters
03:58
04:37 squashable6 left 04:38 squashable6 joined 06:06 sena_kun joined 06:56 domidumont joined 08:12 frost-lab joined 08:44 MasterDuke joined 09:06 Altai-man joined 09:07 Geth left 09:09 sena_kun left 10:55 morayj joined 11:08 squashable6 left 11:11 squashable6 joined
lizmat just pushed 0d1c8a8eefaccf252f7 11:20
11:29 Geth joined 12:45 morayj left 13:01 domidumont1 joined 13:03 domidumont left 13:07 sena_kun joined 13:09 Altai-man left
Geth rakudo: 161325e085 | (Elizabeth Mattijsen)++ | 4 files
Don't use nqp::iterator iterating over mutable QuantHashes

This used to depends on R:I.ITERATIONSET2LISTITER that would create a native string array of the keys of a QuantHash and then return an iterator for it. Changed it to return the native string array instead and adapted the users of it. Also changed the name of the internal method to a more descriptive: IterationSet2keys
Also other assorted cleanup in this area to make better readable code.
13:12
lizmat I think this has now eliminated the use of nqp::iterator on lists in the core
14:18 frost-lab left
timotimo there's no central thing i could override to make something a replacement for ; and i think that's sad 14:19
14:30 leont joined
Geth rakudo: 173205428c | (Elizabeth Mattijsen)++ | src/core.c/Exception.pm6
Fix #4083

Appears that one client of MAYBE-GIST got missed in 973b83de62642a664f17
16:22
linkable6 RAKUDO#4083 [closed]: github.com/rakudo/rakudo/issues/4083 [exceptions] Problem producing message for X::Parameter::RW
16:24 patrickb joined
lizmat timotimo: re "there's no central thing i could override to make something a replacement for ; and i think that's sad", context ? 16:34
Geth rakudo: 0d6278f329 | (Elizabeth Mattijsen)++ | src/core.c/REPL.pm6
Assume Proc as return value as producing output

Fixes #4079. Yes, it is yet another heuristic, but a pretty simple one.
16:40
linkable6 RAKUDO#4079 [closed]: github.com/rakudo/rakudo/issues/4079 shell output pollutes subsequent command output in REPL
Geth rakudo/overridable_semicolon_rule: cc1bac75d8 | (Timo Paulssen)++ | src/Perl6/Grammar.nqp
separate ; out into its own token for easier overriding
17:03
timotimo lizmat: this is the context :)
lizmat ok, I see what you're doing there... but why? 17:05
also: how does this affect compilation times :-)
finally: will it survive rakuast ?
timotimo well, rakuast is free to ignore this change 17:06
17:06 Altai-man joined
timotimo oh, and i should change all those to <.semicolon> actually 17:06
gimme a sec
Geth rakudo/overridable_semicolon_rule: 983a09cb13 | (Timo Paulssen)++ | src/Perl6/Grammar.nqp
separate ; out into its own token for easier overriding
17:07
timotimo there, this won't create the match objects
lizmat fewer Match objects is good! 17:09
17:09 sena_kun left
timotimo well, it'll be the same amount now as before 17:10
sub EXPORT(|) { constant \MAIN = %*LANG<MAIN>; role Amen { token semicolon() { ";" | Amen\. } }; $*LANG.define_slang("MAIN", %*LANG<MAIN> but Amen, %*LANG<MAIN-actions>); {} } 17:11
clearly a very important language extensibility feature 17:12
nine Amazing how long we got by without this 17:18
lizmat pretty sure drforr asked for it at least once, I think 17:19
nine There goes all the beautiful sarcasm 17:20
lizmat sorry... my sarcasm sensor is a little busted
nine Well it is actually amazing that even an obscure feature like this has actual uses :) 17:21
lizmat of course, does it also catch } at the end of a line ending a statement ? 17:22
and does it handle shaped definitions/indexing correctly ? 17:23
17:48 camelCaser joined 17:54 domidumont1 left
Geth rakudo/speedup-native-multidim-array-ops: f27e212bca | (Jonathan Worthington)++ | src/core.c/Rakudo/Internals.pm6
Fix a shaped array performance problem

When we want to turn a native array into a shaped native array, we steal the meta-object and re-use it on a new type with the multi-dim array representation. While that preserves the type behavior (aside from representation), there's a new STable, which lacks the various caches. Having those attached means there's now no non-cached type check or method lookup MOP queries when we work with shaped native arrays, which was the cause of the slow performance compared to non-shaped native arrays.
17:56
rakudo: jnthn++ created pull request #4084:
Fix a shaped array performance problem
lizmat whee! 18:04
18:14 patrickb left
timotimo lizmat: i left out ; from some spots on purpose 18:15
inside argument lists, for example
19:12 domidumont joined 19:16 domidumont left 19:18 Xliff_ joined 19:19 Xliff left
Geth nqp: db6e30309c | (Elizabeth Mattijsen)++ | 2 files
Sort all entries alphabetically for easier lookup

Also add a "sort-ops" script that will make sure that all sections and all ops in "ops.markdown" are in alphabetical order, except for the "Miscellaneous sections" that will always be listed last.
Add a new section is easy: add the section at the end, move any ops that you want in there in that section, and run the "sort-ops" script again. It will automatically add the new section to the table of contents at the beginning.
20:01
cognominal Hi. Building nqp on Apple Silicon fails except when JIT_DISABLE is set to 1. Indeed once the build is done and I unset JIT_DISABLE. lldb says me moar crashes on jit_enter 20:05
Probably the fix is trivial, but I don't know where the config decide to generate code for jitting knowing there is none implemented for ARN64 20:07
lizmat cognominal: probably better asked on #moarvm :-)
or as an issue on MoarVM :-)
cognominal Yes I will write an issue and dig to find a fix 20:09
lizmat ++cognominal
afk for a few hours& 20:10
21:08 sena_kun joined, Altai-man left
dogbert17 cognomial: are you happy with your M1? 21:25
[Tux] Rakudo v2020.11-76-g173205428 (v6.d) on MoarVM 2020.11-54-ga595d9ddc
csv-ip5xs0.810 - 0.815
csv-ip5xs-207.833 - 7.954
csv-parser25.319 - 26.070
csv-test-xs-200.383 - 0.384
test7.720 - 8.070
test-t1.903 - 1.926
test-t --race0.863 - 0.867
test-t-2031.192 - 32.116
test-t-20 --race9.113 - 10.017
21:44
21:50 sena_kun left 22:06 sena_kun joined 22:22 sena_kun left 22:30 bd3i left
Geth roast: vrurg++ created pull request #708:
Add tests for subsets of definites
22:51
rakudo: f27e212bca | (Jonathan Worthington)++ | src/core.c/Rakudo/Internals.pm6
Fix a shaped array performance problem

When we want to turn a native array into a shaped native array, we steal the meta-object and re-use it on a new type with the multi-dim array representation. While that preserves the type behavior (aside from representation), there's a new STable, which lacks the various caches. Having those attached means there's now no non-cached type check or method lookup MOP queries when we work with shaped native arrays, which was the cause of the slow performance compared to non-shaped native arrays.
23:00
rakudo: a89d020757 | (Elizabeth Mattijsen)++ (committed using GitHub Web editor) | src/core.c/Rakudo/Internals.pm6
Merge pull request #4084 from rakudo/speedup-native-multidim-array-ops

Fix a shaped array performance problem
rakudo: ed16d6c109 | (Vadim Belman)++ (committed using GitHub Web editor) | 2 files
Fix handling of definite parameters (#4082)

Restore pre-trsnparentization of nominalizables behavior. Partially fixes #4078.
And a couple of minor improvements.
23:09
linkable6 RAKUDO#4082 [closed]: github.com/rakudo/rakudo/pull/4082 Fix handling of definite parameters
RAKUDO#4078 [closed]: github.com/rakudo/rakudo/issues/4078 [BLOCKER][blin][regression] Blin 2020.12, round 1
Geth roast: 55ee300f69 | (Vadim Belman)++ | S02-types/nominalizables.t
Add tests for subsets of definites

In support of rakudo/rakudo#4082.
Not currently adding a test for declaring an uninitialized variable of type `OfDef` because prior to 6.e a subset was matching itself, event if refinee is a definite.
roast: 806ab82cd7 | (Vadim Belman)++ (committed using GitHub Web editor) | S02-types/nominalizables.t
Merge pull request #708 from vrurg/rakudo_4078

Add tests for subsets of definites
MasterDuke m: my int32 @a[1000000]; my $s = now; @a[$_] = $_ for ^1000000; say now - $s 23:11
camelia 1.06919223
MasterDuke m: my int @a[1000000]; my $s = now; @a[$_] = $_ for ^1000000; say now - $s
camelia 1.058076
MasterDuke committable6: 2020.05 my int32 @a[1000000]; my $s = now; @a[$_] = $_ for ^1000000; say now - $s 23:12
committable6 MasterDuke, ¦2020.05: «5.478343␤»
MasterDuke committable6: 2020.05 my int @a[1000000]; my $s = now; @a[$_] = $_ for ^1000000; say now - $s
committable6 MasterDuke, ¦2020.05: «5.5232091␤»
MasterDuke committable6: 2020.11,HEAD my int @a[1000000]; my $s = now; @a[$_] = $_ for ^1000000; say now - $s 23:17
committable6 MasterDuke, ¦2020.11: «1.0729476␤» ¦HEAD(ed16d6c): «0.09409812␤»
MasterDuke committable6: 2020.11,HEAD my int @a; my $s = now; @a[$_] = $_ for ^1000000; say now - $s
committable6 MasterDuke, ¦2020.11: «0.14525311␤» ¦HEAD(ed16d6c): «0.1245497␤»
MasterDuke jnthn++ 23:18
23:28 vrurg left 23:29 vrurg joined 23:32 squashable6 left 23:33 squashable6 joined
Geth rakudo/rakuast: da88fd5684 | (Jonathan Worthington)++ | 2 files
Correct state var initialization

We need to apply the state initialization check at the point we are doing the assignment of the initializer, not at variable declaration time.
23:35