Perl 6 language and compiler development | Logs at colabti.org/irclogger/irclogger_logs/perl6-dev | For toolchain/installation stuff see #perl6-toolchain | For MoarVM see #moarvm
Set by AlexDaniel on 12 June 2018.
00:00 lizmat joined 00:04 lizmat left 00:10 Kaiepi left 00:37 BenGoldberg joined 02:47 ufobat_ joined 02:50 ufobat left 04:08 xi- left 04:09 xi- joined 05:22 BenGoldberg left 06:28 Kaiepi joined 06:48 robertle joined 06:52 ufobat_ is now known as ufobat
[TuxCM] Rakudo version 2018.06-64-gb659a9c2f - MoarVM version 2018.06-103-gaf455397f
csv-ip5xs0.910 - 0.918
csv-ip5xs-207.405 - 7.732
csv-parser26.379 - 27.225
csv-test-xs-200.467 - 0.470
test9.055 - 9.093
test-t2.303 - 2.347
test-t --race0.966 - 1.063
test-t-2041.875 - 42.239
test-t-20 --race14.855 - 15.403
 07:32
08:06 Ven` joined 09:46 |Tux| joined 10:13 Ven` left 10:15 donaldh joined 10:25 lizmat joined 10:30 BenGoldberg joined 10:32 Ven` joined
tbrowder_ greppable6: caption 10:45
greppable6 tbrowder_, 129 lines, 27 modules: gist.github.com/0701c1ccd6eb261bde...3bde8b4e90
Geth rakudo: 2a414aee18 | (Elizabeth Mattijsen)++ | src/Perl6/Optimizer.nqp
Remove unncessary comment, Ven++
 10:46
tbrowder_ morning all! 10:49
that greppable is very cool!
10:59 Ven` left
Geth rakudo: 5ec2c96ec2 | (Elizabeth Mattijsen)++ | lib/Test.pm6
Fix "diag "foo"|"bar"

We allow junctions to be specified in e.g. "is", but if they fail, they used to bomb on the "diag" being done. This fixes the bombing. This is spectests clean.
Inspired by Text::CSV ecosystem fallout, following hash randomization changes needing tests to be fixed.		 11:21
lizmat m: diag "foo"|"bar" 11:22
camelia 5===SORRY!5=== Error while compiling <tmp>
Undeclared routine:
diag used at line 1
lizmat m: use Test; diag "foo"|"bar" # duh
camelia This type cannot unbox to a native string: P6opaque, Junction
in sub _diag at /home/camelia/rakudo-m-inst-1/share/perl6/sources/C712FE6969F786C9380D643DF17E85D06868219E (Test) line 440
in sub diag at /home/camelia/rakudo-m-inst-1/share/perl6/s…
11:39 Ven` joined
Geth nqp/truffle: 1dd77223dc | (Paweł Murias)++ | 3 files
[truffle] Abstract away call arguments, so that we may pass extra stuff
 11:48
lizmat trivia question: what's so special about 20 September this year ? 11:56
11:59 Zoffix joined
Zoffix lizmat: maybe fix needs to be done somewhere earlier? 'cause right now it has this behaviour: 11:59
c: HEAD use Test; is ^100 .any, 46 12:00
committable6 Zoffix, ¦HEAD(5ec2c96): «ok 1 - ␤»
Zoffix c: HEAD use Test; is ^100 .any, 460
committable6 Zoffix, gist.github.com/18faedb3c0d9e950a8...efbaa1dc7e
Zoffix and instead it should probably print "got: any(…)" or something
lizmat well, this was the least intrusive fix I could think of, apart from full-blown Junction support 12:01
Zoffix c: HEAD use Test; is-deeply ^100 .any, 460
committable6 Zoffix, gist.github.com/04e5afa570818f54f7...e242acaf06
Zoffix OK, I'll just file a ticket
Geth rakudo: e4e753b023 | (Elizabeth Mattijsen)++ | src/core/Cool.pm6
Remove "uniscript"

Clearly this was *way* too naive. See GH#2003		 12:07
synopsebot GH#2003 [open]: github.com/rakudo/rakudo/issues/2003 Should we allow identifiers with different scripts?
Zoffix R#2017
synopsebot R#2017 [open]: github.com/rakudo/rakudo/issues/2017 Test.pm6's Junction support is haphazard
Zoffix lizmat: so what's special about September? 12:11
lizmat answer to trivia question: 20 Sep 2018 is 1000 days after Christmas
Zoffix Ah, cool :)
Perfect time for 6.d to be ready to ship *wink* *wink* :) 12:12
lizmat hehe.... thought has crossed my mind 12:13
it's also like 12 days after the Swiss Perl Workshop 12:14
and not too far from 2018.09 release date 12:15
Geth 6.d-prep: 714262b058 | (Zoffix Znet)++ (committed using GitHub Web editor) | TODO/README.md
Remove `Dismorphia` butterfly name name proposal

  - Someone pointed out it's also a name of some disease
  - Remove my comments about LTAness of Diwali. I no longer care about them and
   butterfly names weren't a widely accepted idea.		 12:20
timotimo .o( people like comma well enough ) 12:21
jnthn wonders how many people (aside from those reading the FAQ) realize it's a butterfly name rather than a ,
Zoffix I didn't know there was a butterfly named comma :P 12:22
MasterDuke neither
jnthn :D 12:23
MasterDuke jnthn, timotimo: is EVAL supposed to be thread-safe? or does the language not guarantee that? 12:24
timotimo it was worth it for me to bring that up
i'd say it should be made threadsafe
MasterDuke pretty sure i looked at the faq, but i guess i missed that
i was hoping ugexe's commit yesterday would have helped, but didn't seem to make a difference 12:25
timotimo en.wikipedia.org/wiki/Polygonia_c-album - Polygonia c-album (comma) is a food generalist (polyphagous) butterfly species belonging to the family Nymphalidae.
MasterDuke at least for `perl6 --profile -e 'say "127.0.0.1" ~~ /^ [ <{ 0 .. 255 }> ]**4 % \. $/; say now - INIT now'`
jnthn MasterDuke: I think the language should guarantee it
Hm, there's no threading in that? 12:26
Zoffix MasterDuke: FWIW, there's more thread unsafeness that ugexe's commit didn't fix: R#1391
synopsebot R#1391 [open]: github.com/rakudo/rakudo/issues/1391 EVAL is not thread safe
Zoffix Well, I guess I don't know for sure if that's the cause of EVAL's crashing 12:28
s/crashing/unsafeness/;
MasterDuke jnthn: github.com/rakudo/rakudo/issues/2013 12:31
isn't the <{ ... }> in the regex causing an EVAL? 12:32
Zoffix MasterDuke: yes, but there's no threading involved
MasterDuke Zoffix: hm, it only crashes when profiling, but i guess that doesn't have to mean threads... 12:33
jnthn No, I expect the profiler crash is something else
MasterDuke actually, didn't timotimo make some recent fix where EVAL or something like that wasn't propagating the is_profiling fact? 12:34
12:34 Zoffix left
timotimo wasn't me 12:35
but if eval is being activated again inside the eval, that would be bad
MasterDuke maybe you were commenting that was a potential cause of a problem
timotimo ah, it's supposed to throw an exception that profiling has already started 12:37
and it doesn't call MVM_profile_start a second time before it segfaults on my end
though the first time it actually ran to where it outputs the profile
12:38 brrt joined
MasterDuke oh right, weren't we were debugging something else and shaved the yak all the way down to seeing if MVM_profile_(start|end) were getting called the correct number of times? 12:39
timotimo right, i seem to recall perhaps profile_end was being called too often 12:40
MasterDuke ah, i think we got onto that because i was trying to --profile-compile the rakudo build and it wasn't outputting anything 12:41
timotimo ah, mhm
fun. the crash is inside a routine that's about 3k instructions long
MasterDuke i don't remember where we left off with the --profile-compile stuff... 12:42
timotimo it's always great when the dump_bytecode output has a -> that's nowhere near the instruction you're looking at, like in this case :| 12:46
MasterDuke right, i'd dumped the bytecode of something, don't remember if you'd found it useful 12:50
timotimo i've got an rr trace now, that means i can step backwards from the crash and just compare what opcodes are being run :)
MasterDuke nice 12:51
timotimo oh, i must recompile with --optimize=0, or else i'll have a hard time 12:56
hm. developing an IDE is a hairy affair, so the butterfly fits there, too :P 13:38
13:38 brrt left 13:41 brrt joined 13:54 brrt left
Geth rakudo: jkramer++ created pull request #2018:
Issue 1906		 14:04
rakudo: jkramer++ created pull request #2019:
Always coerce first argument to &round to Numeric.		 14:07
rakudo: e547811a30 | (Jonas Kramer)++ | src/core/Numeric.pm6
Always coerce first argument to &round to Numeric.

Fixes rakudo/rakudo#1745.		 14:09
rakudo: 2f8074beb2 | (Zoffix Znet)++ (committed using GitHub Web editor) | src/core/Numeric.pm6
Merge pull request #2019 from jkramer/issue-1746

Always coerce first argument to &round to Numeric.
synopsebot RAKUDO#1745 [closed]: github.com/rakudo/rakudo/issues/1745 [consistency][easy to resolve][good first issue] &round inconsistent on types of arguments.
14:31 Zoffix joined
Zoffix ZOFVM: Files=1297, Tests=153433, 172 wallclock secs (27.99 usr 3.66 sys + 3666.80 cusr 190.53 csys = 3888.98 CPU) 14:31
That's a bunch slower than earlier runs :/
14:31 donaldh left
Zoffix Wonder if locking for each serialization context is what's to blame. 14:32
There's a whole ton of tests that use eval to run the test.
jnthn Doubt it; it just takes a lock while it increments an integer once per eval 14:34
If you mean the ugexe++ patch, anyway
Zoffix Yeah, this: github.com/rakudo/rakudo/commit/80...8c54ea026b
But if I have 1000 tests, that's 1000 locks
jnthn m: my $lock = Lock.new; my $i = 0; for 1..100000 { $lock.protect: { $i++ } }; say now - ENTER now 14:35
camelia 0.24730315
jnthn m: my $lock = Lock.new; my $i = 0; for 1..1000000 { $lock.protect: { $i++ } }; say now - ENTER now
camelia 2.2474774
Zoffix ok
m: my atomicint $i; for 1..1000000 { $i⚛++ }; say now - ENTER now 14:36
camelia 0.30206118
Zoffix s: postfix:<⚛++>, \($)
SourceBaby Zoffix, Something's wrong: ␤ERR: ===SORRY!=== Error while compiling -e␤Calling postfix:<⚛++>() will never work with signature of the proto ($, *%)␤at -e:6␤------> put sourcery( <HERE>postfix:<⚛++>, \($) )[1];␤
Zoffix huh
timotimo needs specifically a writable container that's a native int
Zoffix Why is it telling me `Calling postfix:<⚛++>()` when that's not what I called? 14:37
s: postfix:<⚛++>, \(my atomicint $x)
SourceBaby Zoffix, Something's wrong: ␤ERR: ===SORRY!=== Error while compiling -e␤Calling postfix:<⚛++>() will never work with signature of the proto ($, *%)␤at -e:6␤------> put sourcery( <HERE>postfix:<⚛++>, \(my atomicint $x) )[1];␤
Zoffix ah, I forgot the &
s: &postfix:<⚛++>, \(my atomicint $x)
SourceBaby Zoffix, Sauce is at github.com/rakudo/rakudo/blob/2f80...ps.pm6#L81
timotimo aaah
Zoffix So, is it OK to swap that lock to an atomicint or is it too small? 14:38
m: uint32.Range.max.say
camelia 4294967295
Zoffix m: int32.Range.max.say
camelia 2147483647
Zoffix ZOFVM: Files=1297, Tests=153433, 169 wallclock secs (27.00 usr 3.80 sys + 3648.36 cusr 172.61 csys = 3851.77 CPU) 14:39
too small, I think, if it's incremented for each eval 14:43
14:43 brrt joined
Zoffix ZOFSPECTEST: Files=1242, Tests=76420, 130 wallclock secs (18.37 usr 3.36 sys + 2611.22 cusr 152.56 csys = 2785.51 CPU) 14:54
14:55 Zoffix left 15:06 Ven` left 15:10 Ven` joined
AlexDaniel weekly: 15:26
notable6 AlexDaniel, 18 notes: gist.github.com/fe21e524aeac4f17b4...0d82e835c4
AlexDaniel weekly: reset
notable6 AlexDaniel, Moved existing notes to “weekly_2018-07-04T15:26:42Z”
AlexDaniel weekly: leejo.github.io/2018/07/04/five_years_of_spw/
notable6 AlexDaniel, Noted!
lizmat AlexDaniel: ah, good point :-) 15:27
15:28 robertle left 15:31 perlpilot joined
lizmat .tell moritz perl6book.com doesn't show "Parsing with Perl 6 Regexes and Grammars" ? 15:32
yoleaux lizmat: I'll pass your message to moritz.
lizmat .tell moritz well, at least in the graph :-) 15:34
yoleaux lizmat: I'll pass your message to moritz.
Geth rakudo/master: 4 commits pushed by (Jonas Kramer)++, (Zoffix Znet)++ 15:42
roast: jkramer++ created pull request #444:
Tests for rakudo/rakudo#1745.		 15:55
synopsebot RAKUDO#1745 [open]: github.com/rakudo/rakudo/issues/1745 [consistency][easy to resolve][good first issue][testneeded] [RESOLVED; NEED TEST] &round inconsistent on types of arguments.
Geth roast: 955b0e0203 | (Jonas Kramer)++ | S32-num/rounders.t
Add tests for rakudo/rakudo#1745 plus some more for 2-argument version of &round.
 16:05
roast: fc20825797 | (Jonas Kramer)++ | S32-num/rounders.t
Use numeric comparison for &round tests.
roast: 1e17e668ed | (Zoffix Znet)++ (committed using GitHub Web editor) | S32-num/rounders.t
Merge pull request #444 from jkramer/rakudo-1745

Tests for rakudo/rakudo#1745.
synopsebot RAKUDO#1745 [open]: github.com/rakudo/rakudo/issues/1745 [consistency][easy to resolve][good first issue][testneeded] [RESOLVED; NEED TEST] &round inconsistent on types of arguments.
rakudo: 4b0f398537 | (Zoffix Znet)++ | tools/contributors.p6
chmod +x ./tools/contributors.p6
 16:10
16:13 jkramer joined 16:14 AlexDaniel left 16:15 BenGoldberg left
tbrowder_ .tell moritz don’t see that book on perl6.org/resources either 16:27
yoleaux tbrowder_: I'll pass your message to moritz.
16:28 robertle joined 16:31 scovit left 16:39 Ven` left
samcv is wondering how many more modules and code i'll break when i randomize hash's more 16:50
essentially the order of iteration will change whenever you add something to the hash. there's parts in spectest that assume modifying the hash won't cause the order to change
jnthn That's already not reliably true, though, is it? 16:52
(Like if it resizes)
samcv well. true on resizes
also if the table gets resized while you're iterating. that could be very bad
Geth roast: jkramer++ created pull request #445:
Add tests for rakudo/rakudo#1906.		 16:53
synopsebot RAKUDO#1906 [open]: github.com/rakudo/rakudo/issues/1906 [RFC][docs needed][easy to resolve][good first issue][testneeded] [RESOLVED; NEED TEST] &prompt should pass stuff through `val`
samcv so i'm thinking we might want to make it an exception trying to grab another thing from the same iterator
if you've added something to the hash
also there's a test i'm not sure why it fails 16:55
ah nvm i think the test was changed. i'm gonna update my rakudo 16:58
17:06 scovit joined
samcv jnthn: what are your thoughts on making adding entries to a hash a throwable offense? 17:08
or anyone else can chime in
jnthn It's the iteration that'd throw rather than the adding itself, I guess?
Ah, you said so above. Sorry, debugging something and only half reading. :) 17:09
samcv yeah the iteration 17:10
as long as you don't try and get another thing from the iterator you're fine
jnthn What about deletion from the hash? 17:13
samcv you can delete from the hash. though i guess you should only delete the current entry you're on 17:14
if you delete another entry somewhere else that could be bad though
17:22 raschipi joined 17:32 Kaiepi left 17:33 Kaiepi joined 17:55 Kaiepi left 17:56 Kaiepi joined 18:08 lizmat left 18:17 lizmat joined
Geth synopsebot: f03793db5a | (Zoffix Znet)++ | 2 files
Add debug launcher script
 18:26
synopsebot: 6b1a794aee | (Zoffix Znet)++ | dev-start-localhost
Launch correct script
18:50 brrt left
Geth synopsebot: 386197a8d7 | (Zoffix Znet)++ | 7 files
Implement geth-waiter feature

When Issue links are said by Geth, wait for Geth to finish talking
  (currently this commit crashes with weird error; making
a commit to reference it in Rakudo issue)		 18:52
raschipi Thanks Zoffix++ 19:02
Geth synopsebot: f3398f96bb | (Zoffix Znet)++ | 4 files
Fix some compile errors
 19:14
[Coke] ping zoffix 19:20
Geth synopsebot: 3eaf934046 | (Zoffix Znet)++ | 7 files
Revert geth-waiter feature

Doesn't run due to #2023 and I'm too frustrated now to figure out wtf causes that error.		 19:24
19:24 Zoffix joined
Zoffix [Coke]: sup 19:24
19:25 brrt joined
raschipi Zoffix: Thanks anyway. 19:27
19:44 perlpilot left 19:47 Kaiepi left, Kaiepi joined 19:52 ufobat left 20:16 AlexDaniel joined 20:21 dct joined 20:22 j3nnn1 left 20:35 dogbert17 left
Geth ¦ rakudo: zoffixznet self-assigned Missing MoarVM contributors since 2016.09 release github.com/rakudo/rakudo/issues/2024 20:43
Zoffix fus'ake :) I wonder how many people are missing :) 20:49
20:50 j3nnn1 joined 21:04 brrt left 21:10 robertle left 21:12 BenGoldberg joined 21:29 raschipi left 22:15 perlpilot joined 22:27 perlpilot left 23:05 stmuk joined
Geth nqp: jstuder-gh++ created pull request #482:
Display build/commit info on config/make		 23:07
rakudo: jstuder-gh++ created pull request #2026:
Display build/commit info on config/make
23:38 Zoffix left

Powered by Cro and the Raku® Programming Language.

Please report any issues / comments / feature requests as an issue on App::Raku::Log.

Thank you!