00:14 travis-ci joined
travis-ci Rakudo build failed. Tom Browder 'Tweak error message' 00:14
travis-ci.org/rakudo/rakudo/builds/686791219 github.com/rakudo/rakudo/compare/1...defb6c0814
00:14 travis-ci left 01:22 dogbert17 left 02:10 squashable6 left 02:11 squashable6 joined, squashable6 left 02:13 squashable6 joined 02:43 travis-ci joined
travis-ci Rakudo build failed. Tom Browder 'Tweak error message' 02:43
travis-ci.org/rakudo/rakudo/builds/686791219 github.com/rakudo/rakudo/compare/1...defb6c0814
02:43 travis-ci left 04:12 donaldh_ joined, donaldh left 06:17 squashable6 left 06:19 squashable6 joined 06:39 patrickb joined
patrickb .tell rba dev.azure.com/Raku Is now officialy ours. :-) The Microsoft support team was really helpful here. 06:41
tellable6 patrickb, I'll pass your message to rba
rba patrickb: Very cool may you invite me? I will send you the mail as pm. 06:44
tellable6 2020-05-13T12:08:08Z #raku-dev <patrickb> rba Prebuilt releases of rakudo for 2020.05.1 are available for download from rooster.uber.space/patcloud/index....EJLQDZcjbf (as always). Can you upload to rakudo.org? Thank you!
06:55 patrickb left 07:45 Geth left 07:47 tyil left 07:49 Geth joined 07:50 tyilanmenyn joined 08:12 tyilanmenyn is now known as tyil 08:14 jjmerelo joined
jjmerelo Morning 08:14
MasterDuke m: class A {}; say A.^^mro; # heh, guess i'm a little too used to git, i typed this expecting it to do `A.HOW.^mro` 08:27
camelia 5===SORRY!5=== Error while compiling <tmp>
Malformed postfix call
at <tmp>:1
------> 3class A {}; say A.^7ā5^mro; # heh, guess i'm a little too used
08:43 Altai-man_ joined
lizmat Files=1306, Tests=111291, 213 wallclock secs (28.78 usr 8.32 sys + 3003.05 cusr 273.50 csys = 3313.65 CPU) 08:45
09:13 jjmerelo left 09:41 jjmerelo joined
lizmat m: my @a; @a.push for ^100000; say now - INIT now # doing nothing about 8x as slow 09:56
camelia 0.1970097
lizmat m: my @a; @a.push($a) for ^100000; say now - INIT now # as doing something
camelia 5===SORRY!5=== Error while compiling <tmp>
Variable '$a' is not declared. Did you mean '@a'?
at <tmp>:1
------> 3my @a; @a.push(7ā5$a) for ^100000; say now - INIT now #
lizmat m: my @a; @a.push(42) for ^100000; say now - INIT now # as doing something 09:57
camelia 0.0403243
jnthn Probably 'cus it hits the slurpy candidate, no? :)
lizmat ok, maybe about 5x as slow then :-)
yeah, I guess...
but shouldn't the slurpy candidate not be faster seeing as it has nothing to do 09:58
jnthn It has to allocate and populate (ultimately from nothing) the slurpy aggregate
lizmat m: my @a; my @b; @a.append(@b) for ^100000; say now - INIT now
camelia 0.0979625
lizmat m: my @a; my @b = 42; @a.append(@b) for ^100000; say now - INIT now
camelia 0.1305966
lizmat yeah, I guess... :-) 09:59
jnthn Look at a profile, I guess...maybe there's some obvious speedup there
lizmat yeah, will do
jnthn But I suspect it's making an iterator over that empty thing too
I know there's an empty iterator, dunno if it's used in all such cases
10:00 jjmerelo left
Geth rakudo: 2aac854fc4 | (Elizabeth Mattijsen)++ | t/05-messages/02-errors.t
Adapt test for adapted sprintf error message
10:09
10:13 MasterDuke left
Geth rakudo: tbrowder++ created pull request #3692:
Correct test for error msg change
10:14
tbrowder you beat me to it--i will not do a web edit again--sorry :-( 10:17
10:26 sena_kun joined 10:27 Altai-man_ left 10:29 MasterDuke joined
lizmat tbrowder: no, I should have waited a bit longer :-) 10:29
10:37 jjmerelo joined 10:41 MasterDuke left 11:00 jjmerelo left 11:08 japhb left 11:40 jjmerelo joined
Geth rakudo: b50fd8b5ca | (Elizabeth Mattijsen)++ | src/core.c/Array.pm6
Streamline ArrayReificationTarget a bit

  - don't iterate over appending buffer, but shift out of it
  - lose some nqp::stmts for readability
11:41
11:55 AlexDaniel joined 11:56 AlexDaniel left, AlexDaniel joined 11:59 AlexDaniel left 12:05 MasterDuke joined
Geth problem-solving: 375ac6c116 | (Aleks-Daniel Jakimenko-Aleksejev)++ | 2 files
Change default assignees
12:09
12:25 Altai-man_ joined 12:27 sena_kun left 12:49 jjmerelo left
[Coke] I have a MS account if someone wants to invite me to dev.azure.com/Raku 13:44
trying to track down the cro abort trap. build a debug-enabled MoarVM. if I try to attach after I fire off the cro command, I get a gdb error, type "continue", then when cro aborts, gdb shows "exited normally". if I try to run cro through perl6-gdb-m, It seems to just hang. 13:57
Any other suggestions on how to debug? (on OS X)
lizmat: are you able to use cro on OS X? 14:00
timotimo on linux i'd say "run execsnoop against it to see which one is the one that borks" but that's ... linux
lizmat [Coke]: yes, well, it's been a while, but a $project I worked on used Cro, and it worked fine on my MBP 14:01
[Coke] can you try an updated rakudo/cro and do a 'cro stub app app' at some point? I assume it's just me, but... 14:03
I used it many moons ago, also
lizmat [Coke]: will try 14:05
Geth rakudo: 65e17c40ef | (Elizabeth Mattijsen)++ | src/core.c/Array.pm6
Mark two helper methods as private

No idea why there were not private before. Might help a bit in performance.
14:06
[Coke] lizmat++ thank you. 14:07
14:26 sena_kun joined
[Coke] figured I'd do a spectest in this rakudo to see if I get the same issue on something smaller than cro. it tells me inline:p5 isn't installed, tests won't be run. run the command it gives me. rerun spec test. still tells me I:P5 isn't installed. 14:26
14:28 Altai-man_ left 14:29 Kaiepi joined
nine [Coke]: try ./perl6-m -Ilib -e 'use Inline::Perl5' 14:34
should tell you what's wrong
lizmat [Coke]: are you on HEAD? there was an issue a few days ago with dependencies not being recorded correctly 14:37
[Coke] lizmat: it's recent...
lizmat if it's more than 2 days old, you might have that issue
b71cb3ae7c79ea5e9 14:38
linkable6 (2020-05-12) github.com/rakudo/rakudo/commit/b71cb3ae7c Fix main-version.nqp dependencies
lizmat m: dd $*OUT.eof, $*OUT.EOF # why do we have both? 14:42
camelia Bool::False
0
lizmat perhaps EOF should be a private method ? 14:43
[Coke] lizmat: 5561669b3 14:45
nine [Coke]: so get yourself a less ancient rakudo ;) 14:47
[Coke] btw, did get some spectest failures: gist.github.com/coke/ea11adcb1fdcb...b90d36aca8 14:48
rebuilding...
lizmat afk for a few hours& 15:11
15:12 jjmerelo joined
nine No joy trying to reproduce my BEGIN time EVAL issue in a reduced test case so far but....it's not that difficult to provoke other errors 15:23
Ooooh....it's not even a different issue after all. The "chars requires a concrete string, but got null" just happens when trying to report the "missing static code ref for closure" 15:26
MasterDuke ha 15:30
nine So, it really looks like a nested compilation at BEGIN time makes us lose static code refs for closures of the outer compilation unit
Shortest golf now: use NativeCall; sub abort() is native(Str) { !!! }; BEGIN EVAL ''; 15:33
Precompile that and: Serialization Error: missing static code ref for closure 'abort' (:1)
[Coke] the rebuild has no impact on my cro error. 15:34
also no impact on my zef/inline p5 issue. make spectest says it's not installed. zef says it is. 15:35
nine And ./perl6-m -Ilib -e 'use Inline::Perl5'? 15:42
15:43 patrickb joined
patrickb .tell rba Prebuilt releases of rakudo for 2020.05.1 are available for download from rooster.uber.space/patcloud/index....EJLQDZcjbf (as always). Can you upload to rakudo.org? Thank you! 15:44
tellable6 patrickb, I'll pass your message to rba
MasterDuke nine: fwiw i have the same problem as [Coke] with I::P5 and that give me `===SORRY!=== Error while compiling -eCould not find Inline::Perl5 in: <...>` 15:46
[Coke] nine: ./perl6-m dies with: 15:50
Unhandled exception: While looking for '.../sandbox/rakudo/perl6.moarvm': no such file or directory 15:51
(that path edited to remove HOMEDIR)
I'm doing a build in that dir and installing to ~/p6
nine And ./rakudo-m? 15:52
[Coke] can't find I::p5
the suggested install command is to .../sandbox/rakudo/t/../gen/build_rakudo_home/site 15:54
[Coke] opened github.com/croservices/cro/issues/120, even though I'm pretty sure it's not a cro issue. (this way if someone else trips over it, they'll see it in the repo) 15:58
16:25 Altai-man_ joined 16:27 sena_kun left 16:48 jjmerelo left 16:52 Altai-man_ left 16:56 MasterDuke left 17:08 MasterDuke joined 17:16 sena_kun joined 17:27 japhb joined
tbrowder i'm rummaging around Rakudo::Internals and see a list of leap second and the last date listed is in 2016. does that sound right? 17:27
[Coke] tbrowder: yes. 17:29
en.wikipedia.org/wiki/Leap_second - table shows the last one added in 2016
tbrowder thnx, just see that
17:58 jjmerelo joined
lizmat [Coke]: what about 5561669b3 ? 18:17
linkable6 (2020-05-12) github.com/rakudo/rakudo/commit/5561669b3f Streamline variable "is" / "of" handling
[Coke] lizmat: that was the sha I was on. I've already updated to a commit from earlier today, no impact. 18:21
lizmat :-( am in the middle of debuggnig something, will try in half an hour or so
[Coke] no rush or worries
18:25 Altai-man_ joined 18:27 jjmerelo left, sena_kun left
nine Huh...maybe it's not so much that the outer compilation unit is losing a static code ref for a closure and more like somehow the BEGIN time EVAL is causing it to need one where it wouldn't otherwise. 18:36
[Coke] nine: is that on my bug?
nine Dumping the QAST in compile_in_context gives the surprising result of the BEGIN time EVAL somehow causing the complete outer unit to be called in compile_in_context 18:37
[Coke]: no, that's an issue I've been working on for a couple of days
[Coke] cool. 18:38
18:40 Kaiepi left, Kaiepi joined 18:41 Kaiepi left
Geth rakudo: 61046d7695 | (Elizabeth Mattijsen)++ | src/core.c/IO/Handle.pm6
Add IO::Handle.do-not-close-automatically and some reorganization

  - Add .do-not-close-automatically method
   This marks the IO::Handle not being closed by Raku automatically
   either by having DESTROY called on it, or on process exit.
   Returns a Bool indicating success of marking.
  - Add !close method
   This now handles all logic around closing of an IO::Handle, which
   centralizes some logic that was spread around before.
  - Rewrite some internal methods to be more readable
  - Add non-MoarVM versions of do-not-close-automatically and !close
18:47
lizmat Argh, forgot to mention github.com/rakudo/rakudo/pull/3690
18:59 Kaiepi joined 19:13 finsternis left 19:14 donaldh joined 19:15 donaldh_ left
[Tux] Rakudo version 2020.05.1-104-g65e17c40e - MoarVM version 2020.05-12-g3605efdbf
csv-ip5xs0.820 - 0.832
csv-ip5xs-208.229 - 8.229
csv-parser25.425 - 28.137
csv-test-xs-200.379 - 0.389
test7.354 - 7.615
test-t1.868 - 1.872
test-t --race0.828 - 0.918
test-t-2031.561 - 32.496
test-t-20 --race9.032 - 9.389
19:28
Note: I plan to replace my HDD with an SSD this weekend. If I succeed, the timings might change (a lot) 19:29
20:05 donaldh left, donaldh_ joined
moritz and if it doesn't succeed, we won't hear from you for several days? :D 20:10
20:25 sena_kun joined 20:27 Altai-man_ left
nine I think, I'm hot on the trail... 20:49
I managed to get a full backtrace of the superfluous call to the outer unit and it seems to originate at github.com/rakudo/rakudo/blob/mast....nqp#L1298 20:51
lizmat intrigiung 20:53
nine Looks like $*UNIT is the outer unit there, not the EVAL's
lizmat doesn't EVAL set its own $*UNIT ? 20:54
nine It's set in the grammar in token comp_unit
lizmat m: slurp "ten","twenty" 20:55
camelia Too many positionals passed; expected 1 argument but got 2
in block <unit> at <tmp> line 1
lizmat so I want to change this to make it a compile time error
gist.github.com/lizmat/943e0da61d8...d159729594
but this causes a test failure because there is some issue with the candidates 20:56
but I'm not seeing it
nine Oooooh! 20:58
Of course!
Thing is, $*UNIT is declared in the grammar. It starts to parse. It encounters the BEGIN block. The BEGIN block does an EVAL which fires up another compiler which writes to $*UNIT, so when the outer compiler's comp_unit action gets called, there's already a value there 21:00
But wait...the inner compiler declares its own $*UNIT, so it shouldn't touch the outer
lizmat it's really a "my $*UNIT" ? 21:01
nine And it's the inner compiler's comp_unit action that sees the outer's $*UNIT and calls it. Why doesn't it see its own instead?
:my $*UNIT;
lizmat caching ?
or maybe :my $*UNIT is lexical in the wrong scope ? 21:02
nine Even declaring it in EVAL itself doesn't make a difference. I'd guess that someone re-sets it somewhere, but I can't find anything like that. 21:11
Anyway, I think that's very decent progress already for today. More tomorrow :)
Geth rakudo: 22f4344443 | (Elizabeth Mattijsen)++ | 2 files
Tighten up sub slurp / spurt signature

Basically make 'slurp "foo", "bar"' a compile time error, rather than a runtime error.
Sadly, this creates a problem in the proto-arity-count check, of which I have no idea how to fix it. So exclude "slurp" and "spurt" for now from those tests. Fixes welcome!
21:13
21:25 lichtkind joined 21:31 hoelzro left 21:42 hoelzro joined 21:44 Kaiepi left 21:47 Kaiepi joined 21:48 Kaiepi left, Kaiepi joined 21:52 ufobat_ joined
Geth rakudo: b4b2a5ecda | (Elizabeth Mattijsen)++ | src/core.c/Distro.pm6
Set MacOS codename as Distro.desc
21:52
lizmat [Coke] ^^ :-) 21:53
21:56 ufobat__ left 22:01 dogbert17 joined
Geth rakudo: f33ce01c11 | (Elizabeth Mattijsen)++ | src/core.c/IO/Handle.pm6
IO::Handle.PIO is not always nqp::null if not open

This caused a flapper in S32-io/io-cathandle.t . This now appears to be fixed by checking with nqp::defined, rather than nqp::isnull.
22:03
lizmat sleep& 22:06
22:12 lichtkind left 22:25 Altai-man_ joined 22:27 sena_kun left 22:44 Altai-man_ left 23:12 squashable6 left 23:14 squashable6 joined 23:48 patrickz joined 23:52 patrickb left