00:06 Xliff joined
Xliff o/ 00:11
tellable6 2020-01-22T10:42:54Z #raku <Doc_Holliwood> Xliff: Thanks, but that's not what I meant.
TreyHarris the light dawns... he's asking for Raku to intentionally provoke, and then recover from, a segmentation fault 00:32
for... some reason that's very important to him 00:33
00:45 sena_kun left 00:54 lucasb left 00:59 sena_kun joined 01:03 Xliff left 02:44 sena_kun left 02:59 sena_kun joined
TreyHarris why is .^name preferred to .WHAT in the docs? At least, there's a test to warn against using WHAT and not .^name 03:08
AlexDaniel TreyHarris: that'd be a nice tellable feature, yes 03:32
04:46 linkable6 left, evalable6 left, linkable6 joined, sena_kun left 04:48 evalable6 joined 04:50 AlexDaniel left 05:00 sena_kun joined 06:01 squashable6 left 06:03 squashable6 joined 06:45 sena_kun left 07:00 sena_kun joined 07:34 domidumont joined 07:38 epony left 08:45 sena_kun left 09:01 sena_kun joined
jnthn TreyHarris: I think 'cus if folks stringify the type object it's a warning, so `say $foo.WHAT` or `note $foo.WHAT` works out well, but things like `put $foo.WHAT` or `say "the type is {$foo.WHAT}"` do not 09:43
|Tux| Rakudo version 2020.01-38-gd18c69b2b - MoarVM version 2020.01.1-2-gcbb04d462
csv-ip5xs0.692 - 0.699
csv-ip5xs-205.820 - 6.038
csv-parser22.351 - 22.514
csv-test-xs-200.367 - 0.370
test7.050 - 7.208
test-t1.787 - 1.794
test-t --race0.787 - 0.803
test-t-2030.409 - 30.590
test-t-20 --race8.445 - 8.515
10:32
10:45 sena_kun left 10:59 sena_kun joined 11:14 epony joined
Geth rakudo: b777d6ff61 | (Elizabeth Mattijsen)++ | src/core.c/Regex.pm6
Regex.ACCEPTS should not assume $/ is a container

Since github.com/rakudo/rakudo/commit/d8e859d000 an nqp::istype is done during creation of a BUILDPLAN for a class. In some cases, specifically when the constraint on an attribute is a subtype, this meant calling Regex.ACCEPTS at compile time. However, at that point in time, the nearest $/ is an immutable ... (8 more lines)
11:19
lizmat *phew* that took some debugging... 11:20
but was very clear once I spotted the hidden "=" in the code :-)
11:28 domidumont left 12:45 sena_kun left 12:47 lucasb joined 12:48 Kaiepi left 12:49 Kaiepi joined 13:01 sena_kun joined
Geth nqp: 57f8d2b123 | (Elizabeth Mattijsen)++ | src/QRegex/Cursor.nqp
Only do assignment if nfa is not in the cache

This method is at 25% of --profile-compile -e ''
13:10
13:16 titsuki joined
Geth nqp: 9a12c17c16 | (Elizabeth Mattijsen)++ | src/QRegex/Cursor.nqp
cache_add returns the value, so no assignment needed
13:21
13:24 Kaiepi left 13:30 domidumont joined, domidumont left 13:31 domidumont joined 13:40 titsuki left 13:44 domidumont left 14:00 titsuki joined 14:01 Kaiepi joined 14:02 domidumont joined
Geth nqp: fad1aa45f3 | (Elizabeth Mattijsen)++ | src/QRegex/NFA.nqp
Completely disable NFA debugging (for now)

To gauge its effects on parsing Raku source.
14:04
dogbert11 lizmat: curious wants to know :) does NFA debugging have a noticable impact? 14:12
Geth roast: 355b01d9bd | (Ben Davies)++ | 3 files
Move IO::Socket::INET UNIX socket tests to S32-io/IO-Socket-INET-UNIX.t

UNIX socket support doesn't exist on the JVM or JS backends at the moment, so don't try to test them with them.
14:13
lizmat dogbert1: not sure yet
but it was doing 2 extra calls to a sub in just about every method in that class
even with debugging off
dogbert11 oops
lizmat I'll be afk for a few hours... in the meantime it will run a spectest that should give some more info, I'd say 14:15
afk&
dogbert11 interesting 14:17
[Coke] TreyHarris: (build log) ping jj. I'm out of the loop 14:19
14:19 travis-ci joined
travis-ci NQP build failed. Elizabeth Mattijsen 'Completely disable NFA debugging (for now) 14:19
travis-ci.org/perl6/nqp/builds/643391927 github.com/perl6/nqp/compare/9a12c...d1aa45f305
14:19 travis-ci left
vrurg lizmat: Not only you fixed URI, but I also closed an old issue of mine. Thank you! 14:27
14:45 sena_kun left 14:59 sena_kun joined 15:19 titsuki left
tony-o_ AlexDaniel`: that repo is building raku on an environment similar to travis-ci so you can clone it and use it on travis without waiting for raku to compile or using docker 15:49
tellable6 2020-01-23T22:34:10Z #raku-dev <patrickb> tony-o Is there any write up or source for how zeco works with git repos? I'd be interested how it deals with versions of modules. 15:50
tony-o_ .tell patrickb yes - i can write that up, i'll probably write something up friday or after i get some buy in from ugexe on how we're going to consume this huge meta file
tellable6 tony-o_, I'll pass your message to patrickb
TreyHarris jnthn: That makes sense... WHAT is the type object, so before assignment, $x === $x.WHAT, meaning you must refuse under strict to implicitly val-ify a type object, otherwise strict becomes meaningless 16:25
m: no strict; say "{<1/2>.WHAT}" 16:26
camelia Use of uninitialized value of type Rat in string context.
Methods .^name, .raku, .gist, or .say can be used to stringify it to something meaningful.

in block <unit> at <tmp> line 1
TreyHarris Hmm, not just strict
I guess printing "Rat" when someone expects to be printing an actual number is just too surprising, so it fails even not strict 16:27
Oh right, I forgot that was one of the first ground-shaking Apocalyptic changes, that use of unitialized variables would become fatal rather than warning. Is there a pragma to revert to warning? I imagine not 16:29
tobs TreyHarris: I think it *is* a warning 16:35
m: say "{<1/2>.WHAT}"; say "still alive"
camelia Use of uninitialized value of type Rat in string context.
Methods .^name, .raku, .gist, or .say can be used to stringify it to something meaningful.

still alive
in block <unit> at <tmp> line 1
TreyHarris tobs: oh... what's going on with my REPL?
tobs you just didn't see anything but the warning at first because that's what the warning was about :)
TreyHarris No, I meant my local REPL 16:36
It's definitely throwing and catching exceptions
Oh, dear... I've ruined it by handing it that DWord uint32 abomination
tobs huh, I still have Rakudo 2019.07 and it just warns 16:37
ah, I see.
TreyHarris It's throwing errors for any variables that touch Int now
restart.
tobs Native types do that to the REPL IIRC
TreyHarris Why is Rat affected, I wonder?
I got `Cannot find method 'qast' on object of type NQPMu` using a Rat 16:38
doesn't matter. GIGO
16:44 domidumont left 16:45 sena_kun left 17:01 sena_kun joined
Geth problem-solving/front-matter-for-routines: 4982b8254f | (JJ Merelo)++ | 2 files
Adds proposal for solving #155
17:05
problem-solving: JJ++ created pull request #156:
Adds proposal for solving #155
17:07
17:11 pyrimidine joined 17:14 pyrimidi_ left, patrickb joined
Geth problem-solving/front-matter-for-routines: a0ed4da650 | (JJ Merelo)++ | solutions/documentation/front-matter-for-routines.md
might → will
17:30
17:45 Kaiepi left, Kaiepi joined
TreyHarris m: say so Int.does(UInt); say so UInt.does(Int); # what's going on here? 18:15
camelia True
True
TreyHarris oh, it's dynamic 18:16
m: my Int $y = -3; say so $y.does(UInt); $y = 3; say so $y.does(UInt); 18:18
camelia False
True
TreyHarris duh, block-where clauses make a static analysis equivalent to the halting problem, doesn't it? :-) 18:41
lizmat bisectable6: dd -> $ where /./, $/ {}("x", "y") 18:43
bisectable6 lizmat, Bisecting by exit code (old=2015.12 new=b777d6f). Old exit code: 1
lizmat, bisect log: gist.github.com/8615dc56a3b36ae9cb...a2878078b8
lizmat, (2020-01-29) github.com/rakudo/rakudo/commit/b7...55b5e50813
18:46 sena_kun left
lizmat bisectable6: $/ := 42; dd /./.ACCEPTS("a") 18:47
bisectable6 lizmat, Bisecting by exit code (old=2015.12 new=b777d6f). Old exit code: 1
lizmat, bisect log: gist.github.com/a590b13bf7bc565871...0bf67855ad 18:48
lizmat, (2020-01-29) github.com/rakudo/rakudo/commit/b7...55b5e50813
Geth ¦ problem-solving: Kaiepi assigned to jnthn Issue IO::Socket::INET's methods for socket creation may need reconsideration github.com/Raku/problem-solving/issues/157
roast: 38dbffdb91 | (Elizabeth Mattijsen)++ | S05-mass/rx.t
Add test for GH#2059
18:53
linkable6 GH#2059 [closed]: github.com/rakudo/rakudo/issues/2059 [LTA] LTA error with regex in `where` clause while `$/` is used for a later param in the same signature
18:55 vrurg left 18:56 vrurg joined 19:00 sena_kun joined
Geth rakudo: fb13c31c44 | (Elizabeth Mattijsen)++ | tools/templates/NQP_REVISION
Bump NQP to get some grammar micro-opts

They're little ones for now, but every little bit helps
19:18
rakudo: 59cf387d44 | (Christian Bartolomäus)++ | src/vm/jvm/runtime/org/perl6/rakudo/RakOps.java
[JVM] Update overlooked ref to HLL language perl6

This was done in all other places with b787095358.
19:23
lizmat bartolin++ # thanks for the spot! 19:24
bartolin yw. I'm trying to catch up with recent changes :) 19:25
ah, well. I guess this wants to be changed, too: github.com/rakudo/rakudo/blob/59cf...java#L1003 19:47
bartolin could do that later 19:49
lizmat bartolin: yeah, feels like 19:56
20:01 MasterDuke joined
Geth rakudo: dumarchie++ created pull request #3447:
Merge Iterable and Seq infix:<eqv> multis
20:04
MasterDuke anyone have an objection to merging github.com/MoarVM/MoarVM/pull/1221 ? 20:08
lizmat MasterDuke: I see some unresolved conversations ? 20:12
MasterDuke not anymore! 20:13
lizmat can you merge ?
20:13 dumarchie joined
lizmat m: class A { method !a() { } }; dd A.^find_private_method("a").head.signature 20:14
camelia :(A: *%_)
MasterDuke yeah
lizmat MasterDuke: then by all means,,, go ahead I would say
m: class A { method !a() { } }; dd A.^find_private_method("a").head.signature 20:15
camelia :(A: *%_)
lizmat so why do private methods also get *%_ in their signature ??
their supposed to be more like subs
MasterDuke ok, will do. there's a bunch of further optimization and such, but i'll do those are separate prs 20:16
lizmat m: sub aO() { }; ddd &a.signature
camelia 5===SORRY!5=== Error while compiling <tmp>
Undeclared routines:
a used at line 1
ddd used at line 1. Did you mean 'dd'?
lizmat m: sub a() { }; ddd &a.signature
camelia 5===SORRY!5=== Error while compiling <tmp>
Undeclared routine:
ddd used at line 1. Did you mean 'dd'?
lizmat m: sub a() { }; dd &a.signature
camelia :()
[Coke] Is there any interest in getting rakudo targetting .NET core? 20:29
I have, since the time I switched from a place where running on the jvm would have been very helpful if we wanted to use Raku for work, to a place where running it on .NET would now be helpful. :) 20:30
TreyHarris I just tried to build rakudo manually for the first time in awhile. I ran the literal example in README: `make t/spec/S12-traits/parameterized.t` and it failed with "Dubious, test returned 1 (wstat 256, 0x100)" and no other output. So did basic.t in the same dir. But `make t/01-sanity/*` and `make t/02-rakudo/*` all passed 20:33
MasterDuke [Coke]: how long ago did Niecza build? 20:36
oh, but that wasn't a rakudo backend? it was it's own thing? 20:37
dumarchie TreyHarris: `cd t\spec` and `perl fudgeandrun S12-traits\parameterized.t` 20:41
TreyHarris dumarchie: that gave me gist.github.com/treyharris/09ce0b0...a3c2ed441a 20:43
oops, that doesn't have the stderr, let me try again 20:44
dumarchie: here gist.github.com/db7dd6440564ef3c8f...e257fce2e7
[Coke] MasterDuke: it was its own thing, and that was targeting mono. 20:45
20:45 sena_kun left
dumarchie TreyHarris: wait... 20:46
[Coke] it also has a bootstrap bytecode block, not sure the bootstrappy stuff still works.
20:46 Kaeipi joined, Kaiepi left
MasterDuke hm, so probably none of the code can be reused? 20:49
lizmat no, one would need to boot NQP on .NET to make that work
dumarchie TreyHarris: did you get the warning "did not find local perl6 binary; switching to using system `perl6`"? 20:51
TreyHarris dumarchie: that latter gist is the entire output (I ran `./fudgeandrun S12-traits/parameterized.t |& gist-paste -`) 20:53
and searching scrollback, no, I never got that error
dumarchie Hmm, I got this to work not so long ago 20:54
21:00 sena_kun joined
TreyHarris dumarchie: I'm only concerned whether this is a safe rakudo to use or whether these trait test failures indicate I have a bad build 21:02
dumarchie TreyHarris: I get similar errors on my attempts to run that test, but `gmake spectest` gives me only the usual Windows failures 21:03
Geth rakudo/cci-precomp-build: 5 commits pushed by (Patrick Böker)++ 21:04
vrurg lizmat: I have to leave now, didn't get enough time to check, but I have bad feeling about handling 'will build' by the compile time typecheck. It looks to me that the code typechecks agains the closure/routine, not the value it returns. Though I never used 'will' before and not sure what it has to be. 21:05
lizmat: I plan to check thoroughly later today. 21:06
lizmat I haven't implemented anything wrt "will build" ? I've implemented "is built"
and if you currently use "is built(:bind)" it doesn't typecheck at all atm
except for default values :-) 21:07
vrurg lizmat: the block at line 125 of BUILDPLAN.nqp is where the potential problem is. 'my $default := $_.build;' confuses me because set_build accepts code. That's all I have for now. 21:11
vrurg is afk&
lizmat .tell vrurg but that piece of code has been there for years :-)
tellable6 lizmat, I'll pass your message to vrurg
bartolin TreyHarris: if I'm not mistaken, S12-traits/parameterized.t isn't part of spectest (spectest.data from roast: github.com/perl6/roast/blob/master...est.data).
TreyHarris bartolin: ah... It seems like a strange one to include as an example in the Rakudo build instructions, then 21:12
bartolin oops :/
TreyHarris I dropped a rakudo issue on this 21:20
bartolin TreyHarris++ 21:21
21:21 Kaeipi left 21:26 Kaiepi joined
Geth rakudo: ab6710e605 | (Christian Bartolomäus)++ | README.md
Use file from roast's spectest.data as example

As reported by TreyHarris++ the tests from S12-traits/parameterized.t don't pass. That file is a bad choice for an example anyway, since it is not part of spectest.data from roast. S03-operators/comparison.t is and all tests from that file pass.
21:30
rakudo: 5984ce9813 | (Christian Bartolomäus)++ | src/vm/jvm/runtime/org/perl6/rakudo/RakudoJavaInterop.java
[JVM] Update another ref to HLL language perl6

This was done in all other places with b787095358.
21:44
22:07 dumarchie left 22:46 sena_kun left 22:58 Kaeipi joined 22:59 Kaiepi left 23:02 sena_kun joined 23:08 titsuki joined 23:09 MasterDuke left
Geth rakudo: vrurg++ created pull request #3449:
Typecheck Attribute's build attribute against Code
23:18
vrurg .tell lizmat *That* piece of code was there, but the following lines of the block are mostly yours. ;) Please, merge #3449 unless you have another idea as to how to fix it. 23:24
tellable6 vrurg, I'll pass your message to lizmat
Geth rakudo: a8bb85a007 | (Vadim Belman)++ | src/Perl6/Metamodel/BUILDPLAN.nqp
Typecheck Attribute's build attribute against Code

Attrbiute `build` of `Attrbiute` class can not only be set to a method but to any kind of generic code. This fixes a breakage in `Red` ORM.
23:42
rakudo: e27811a62b | (Elizabeth Mattijsen)++ (committed using GitHub Web editor) | src/Perl6/Metamodel/BUILDPLAN.nqp
Merge pull request #3449 from vrurg/will-build-typecheck

Typecheck Attribute's build attribute against Code
vrurg lizmat: It was a bit preliminary as I just realized that some spectests are actually broken. :( 23:43
lizmat looked ok to me...
tellable6 2020-01-29T23:24:33Z #raku-dev <vrurg> lizmat *That* piece of code was there, but the following lines of the block are mostly yours. ;) Please, merge #3449 unless you have another idea as to how to fix it.
vrurg I'll re-check on clean master. Could be that my repo is polluted with garbage code I overlooked. 23:44
lizmat ah, indeed, breakage :-( 23:46
Cannot find method 'package' on object of type NQPMu 23:47
vrurg Like it can't find Code which is weird. 23:49
23:51 patrickb left
Geth rakudo: b6fa5f5dd1 | (Elizabeth Mattijsen)++ | src/Perl6/Metamodel/BUILDPLAN.nqp
Revert "Typecheck Attribute's build attribute against Code"

This reverts commit a8bb85a0074b4cda9c038bb68243cc664acb7b8d.
Appears this is not the right way.
23:54
lizmat vrurg: is there a ticket for the Red breakage? If so, I will look at it tomorrow
first some sleep&
vrurg lizmat: The PR has a link