Zoffix .ask jnthn any tips for fixing this bug: `m: say &min ~~ Callable; #`(False) &min.^compose; say &min ~~ Callable; #`(True)` Some new commits went in that tickle it and that's currently marked as a release blocker. I can only think of something nutty like collecting all the traited subs and then .^composing it at the end of setting 00:15
yoleaux Zoffix: I'll pass your message to jnthn.
Zoffix .tell jnthn For context, the breakage is caused by `is pure` mixing in a role before Callable is fully composed. It's github.com/rakudo/rakudo/issues/1566 00:16
yoleaux Zoffix: I'll pass your message to jnthn.
Zoffix .tell jnthn and nother hack I can think of is mixing in `role is-pure does Callable` instead of just `role is-pure`. Dunno how much evil it is to have a role (Callable) mixed in from two places like this is; `say &min.^roles; # ((is-pure) (Callable) (Callable))` 00:21
yoleaux Zoffix: I'll pass your message to jnthn.
timotimo it'd probably be fine to put all the "is pure" mixins after Callable is finished 00:25
Geth rakudo: 67f36e3605 | (Timo Paulssen)++ | src/core/IO/Socket/Async.pm6
first draft of "datagram"-oriented API for udp sockets

ask the .Supply to give you :datagrams and you'll receive instances of IO::Socket::Async::Datagram with
  .data being the string or buffer of data and
  .hostname and .port being the sender's address.
rakudo: b406b32006 | (Timo Paulssen)++ | src/core/IO/Socket/Async.pm6
change :datagrams to :datagram, add encode method

and throw exceptions for decode or encode on a datagram with the wrong type of data.
rakudo: ada26ca18e | (Timo Paulssen)++ | src/core/IO/Socket/Async.pm6
Merge branch 'udp_receive_hostname_port'

by passing :datagram to the Supply method on a IO::Socket::Async in udp mode, you can get the hostname and port of the sending socket with every message.
timotimo zscript is testing for the bumpage 00:29
fish: “env ZSCRIPT_DIR=/home/timo/perl…” terminated by signal SIGSEGV (Address boundary error) 00:31
well that isn't fun :o
Geth rakudo: dd2c901921 | (Timo Paulssen)++ | tools/build/NQP_REVISION
[NQP Bump] Brings 3 commits

NQP bump brought: github.com/perl6/nqp/compare/2018....gaa765773d aa765773d [MoarVM Bump] Brings 2 commits 070ec18bb speed up sql output by about 30% c855b116e remove space for consistent format
MoarVM bump brought: github.com/MoarVM/MoarVM/compare/2...g408555456 408555456 ignore libuv-caused "fake" empty datagrams 6466cd639 expose both hostname and port of received datagrams
¦ rakudo: version bump brought these changes: github.com/perl6/nqp/compare/2018....gaa765773d
Zoffix Based on the comment on the script lizmat++ already tried that: "I tried fixing this by moving Routine.pm to before traits.pm so that Routine would be fully composed. But this turned out to be a rabbit-hole that will take some real re-organizing the setting for that to work, as it would also need Attribute.pm before Routine.pm and that is really a can of worms." 00:34
It's not just `is pure`, it's all the traits that get mixed into callables (like `is nodal`, `is default`, `is hidden-from-USAGE` if that's used anywhere) 00:37
I guess lizmat tried rearranging source files of traits not the source of where the traits are used
Zoffix tries that
timotimo: any idea how to find where Callable is finished? None of the `is pure`s are before `src/core/Callable.pm6` in the core files list, so I guess it's not finished there 00:40
timotimo o :(
Zoffix hm, maybe it's not callable, but Routine or something like that. Gonna try puttin all the trait use after all the files for Callable types 00:42
\o/ 00:48
MasterDuke did a bump get messed up? i just built moar and nqp at HEAD, but rakudo is complaining that nqp is too old 00:58
yoleaux 5 Mar 2018 02:33Z <Zoffix> MasterDuke: FWIW rindolf's program in NQP version; still 1.86x slower. Maybe some of these ops don't get JITted? No idea how to check: gist.github.com/zoffixznet/cb19c86...19f7ee9383
5 Mar 2018 07:41Z <brrt> MasterDuke: yeah I'm seeing it. I really wonder how you're getting a CAST_LOAD_ADDR from that
Zoffix FWIW that gist is outdated. Latest version is 2.5x faster than P5 for me 00:59
this one gist.github.com/zoffixznet/4df7a30...fb65bbb0a4 01:00
MasterDuke: probably.... I don't see an nqp push above. 01:01
timotimo aha!
Geth nqp: 070ec18bb2 | (Timo Paulssen)++ | src/vm/moar/HLL/Backend.nqp
speed up sql output by about 30%

by accumulating pieces of string and writing data out to the file when there's enough. Before this change we'd write mostly below 100 bytes per write call, now the number is between 10_000 and 15_000.
Zoffix :)
Geth nqp: aa765773d8 | (Timo Paulssen)++ | tools/build/MOAR_REVISION
[MoarVM Bump] Brings 2 commits

MoarVM bump brought: github.com/MoarVM/MoarVM/compare/2...g408555456 408555456 ignore libuv-caused "fake" empty datagrams 6466cd639 expose both hostname and port of received datagrams
¦ nqp: version bump brought these changes: github.com/MoarVM/MoarVM/compare/2...g408555456
timotimo oh, i wasn't sure if i should push that commit yet, the other one 01:02
oh well, i'll push any necessary debugging afterwards
MasterDuke cool, working now 01:03
timotimo sorry about that 01:04
travis-ci Rakudo build failed. Timo Paulssen 'Merge branch 'udp_receive_hostname_port' 01:09
travis-ci.org/rakudo/rakudo/builds/350095318 github.com/rakudo/rakudo/compare/8...a26ca18e08
MasterDuke timotimo: np. btw, any idea if it would be faster (or just easier on the GC) to create a fixed size nqp::list_s upfront and just stick the bits in? 01:12
timotimo i think the GC only cares about the amount of active slots inside the list; the slot array itself is alloced and realloced to grow; pre-sizing it to the correct size can help, though. 01:14
however, after the list has been full the first time, it won't have to grow any more unless it passes that point again
MasterDuke right, just realized that
but still, it'd be a bunch of bindpos_s instead of push_s, would that make a difference? 01:15
timotimo i would assume push_s to be better since you don't have to carry around the index yourself and do the increments 01:16
Zoffix m: <a b c>.grep: { state %code-files is SetHash = <foo bar>; LEAVE %code-files{$_}--; } 01:23
camelia Invocant of method 'SET-SELF' must be an object instance of type 'QuantHash', not a type object of type 'SetHash'. Did you forget a '.new'?
in block at <tmp> line 1
in block <unit> at <tmp> line 1
Zoffix boo
m: <a b c>.grep: { state %code-files is SetHash = <foo bar>; LEAVE dd %code-files }
camelia SetHash.new("foo","bar")
Zoffix `state` is buggy AF
MasterDuke hm, if my benchmark is representative, bindpos_s is faster and uses less memory 01:24
timotimo oh? i'd be interested in that benchmark 01:25
Zoffix Ah, it's RT#131508 01:26
synopsebot RT#131508 [new]: rt.perl.org/Ticket/Display.html?id=131508 [BUG] `state` with % is Sethash in whenever block ends up with a type object on second iteration
MasterDuke 'my int $a := 0; my $l := nqp::list_s(); nqp::setelems($l, 500); while $a++ < 1_000_000 { my int $i := 0; while $i++ < 500 { nqp::bindpos_s($l, $i, "asdfasdfasdfasdfasdfasdf") } }'
vs 'my int $a := 0; my $l := nqp::list_s(); while $a++ < 1_000_000 { my int $i := 0; while $i++ < 500 { nqp::push_s($l, "asdfasdfasdfasdfasdfasdf") } }'
/usr/bin/time reported 11.4s for push_s and 3926972maxresident 01:27
but 7.8s for bindpos_s and 20536maxresident
oh, whoops, forgot to clear the array in the push_s version
travis-ci Rakudo build passed. Timo Paulssen 'Merge branch 'udp_receive_hostname_port'
travis-ci.org/rakudo/rakudo/builds/350095318 github.com/rakudo/rakudo/compare/8...a26ca18e08
timotimo fascinating 01:28
warrants a closer look i'd say
but i'll go sleep first
MasterDuke add the splice is a little better 01:29
'my int $a := 0; my $l := nqp::list_s(); my $e := nqp::list_s(); while $a++ < 1_000_000 { my int $i := 0; while $i++ < 500 { nqp::push_s($l, "asdfasdfasdfasdfasdfasdf") }; nqp::splice($l, $e, 0, 500) }'
10.3s and 20948maxresident 01:30
so mem is the same as bindpos_s, but it's a little slower
timotimo oh, haha 01:31
yeah, you weren't clearing out the list between the runs
so the push version would make a million more entries into that hash
sorry, list
Zoffix m: say Mu.^lookup('splice') ~~ Callable 01:32
camelia False
Zoffix curses... I can move routines, but not the classes
And I already wrote a test covering the use -_-
m: Mu.^lookup("splice").^compose; say Mu.^lookup('splice') ~~ Callable 01:34
camelia True
Zoffix guess I'll hack it that way
grrrrr 01:39
oh nm, it might be OK. But there's `HYPER( sub (\obj) is nodal { obj."$meth-name"(|c) }, SELF ),` in Mu.pm6
travis-ci Rakudo build failed. Timo Paulssen '[NQP Bump] Brings 3 commits 01:41
travis-ci.org/rakudo/rakudo/builds/350096653 github.com/rakudo/rakudo/compare/a...2c901921b4
Zoffix So.... wtf is there a proto-only for splice in Mu?
m: Any.^lookup('EXISTS-KEY').^compose; say Iterator.^lookup('EXISTS-KEY') ~~ Callable 01:48
camelia False
Zoffix ZofBot: cue expletives
ZofBot Zoffix, And Father said he would pay someone £50 to do it after school and he wasn’t going to take no for an answer
Zoffix Right this ain't gonna work. 01:49
Geth rakudo/fix-Callable-composition: 3506395576 | (Zoffix Znet)++ | 6 files
Draft of fixing some routines not doing Callable role

Was working out OK, until hitting this issue: Any.^lookup('EXISTS-KEY').^compose; say Iterator.^lookup('EXISTS-KEY') ~~ Callable; # False
So not only we need to recompose methods, but also need to go through kids and recompose them...
Zoffix .tell jnthn Another idea that promised to be OK, but ran into hacky category (mentioned in commit message): github.com/rakudo/rakudo/commit/3506395576 The start of src/core/traited-routines.pm6 file added there does .^compose on methods [missing BEGIN block]. Is it too much do go through all methods of all types defined by that point and do .^compose on anything that !~~ Callable? 01:53
yoleaux Zoffix: I'll pass your message to jnthn.
Geth rakudo: 3259953e5f | usev6++ | t/spectest.data
Don't run stress tests for lock-async on JVM

Since fixing the ThreadPoolScheduler (commit 8d954027fd) these tests
  are dying badly. I'm quite sure, an underlying problem is surfacing
travis-ci Rakudo build passed. usev6 'Don't run stress tests for lock-async on JVM 07:30
travis-ci.org/rakudo/rakudo/builds/350179339 github.com/rakudo/rakudo/compare/d...59953e5f65
Geth roast: 2592631616 | LLFourn++ | S03-operators/set_elem.t
Tests for GH #1593

|Tux| Rakudo version 2018.02.1-129-g3259953e5 - MoarVM version 2018.02-33-ge639691a7
csv-ip5xs0.795 - 0.799
csv-ip5xs-207.455 - 7.480
csv-parser36.070 - 37.049
csv-test-xs-200.430 - 0.468
test8.952 - 9.056
test-t2.596 - 2.619
test-t --race1.048 - 1.054
test-t-2045.727 - 46.876
test-t-20 --race15.769 - 15.781
travis-ci Rakudo build passed. Timo Paulssen '[NQP Bump] Brings 3 commits 11:56
travis-ci.org/rakudo/rakudo/builds/350096653 github.com/rakudo/rakudo/compare/a...2c901921b4
Geth nqp: 7e8e93012a | (Tom Browder)++ (committed using GitHub Web editor) | README.pod
add reference to built-in routines
dogbert17 multi sub foo(Int $x where 1..10) { state $bar = 42; say $bar; }; foo(3); 12:16
evalable6 (Any)
dogbert17 multi sub foo(Int $x) { state $bar = 42; say $bar; }; foo(3);
evalable6 42
timotimo huh, that's fascinating 12:18
dogbert17 timotimo: RT #130257 12:19
synopsebot RT#130257 [new]: rt.perl.org/Ticket/Display.html?id=130257 [BUG] state variable not initialized in some multi subs
Geth nqp: 5bc5283693 | (Tom Browder)++ (committed using GitHub Web editor) | examples/fib.nqp
don’t need “nqp::” for built-ins
nqp: ffa648a2fe | (Tom Browder)++ (committed using GitHub Web editor) | examples/loops.nqp
make example show off nqp features a bit more
nqp: ea1349b348 | (Tom Browder)++ (committed using GitHub Web editor) | README.pod
add note about speed-ups using NQP
travis-ci NQP build failed. Tom Browder 'add note about speed-ups using NQP' 12:52
travis-ci.org/perl6/nqp/builds/350295474 github.com/perl6/nqp/compare/ffa64...1349b3486c
Geth nqp: a065e99f08 | (Tom Browder)++ | 2 files
rename and prettify with markdown
nqp: 17a1eedeae | (Tom Browder)++ | 8 files
Merge branch 'master' of github.com/perl6/nqp
travis-ci NQP build passed. Tom Browder 'Merge branch 'master' of github.com/perl6/nqp' 13:18
travis-ci.org/perl6/nqp/builds/350305437 github.com/perl6/nqp/compare/ea134...a1eedeae06
Geth nqp: 83469703e5 | (Zoffix Znet)++ | tools/build/MOAR_REVISION
[MoarVM Bump] Brings 2 commits

MoarVM bump brought: github.com/MoarVM/MoarVM/compare/2...3-ge639691 e639691 [JIT] Release only spill slots without active overlap dffe6f9 We already have a MVM_spesh_get_lex_type
¦ nqp: version bump brought these changes: github.com/MoarVM/MoarVM/compare/2...3-ge639691
rakudo: ec9a227b37 | (Zoffix Znet)++ | tools/build/NQP_REVISION
[NQP Bump] Brings 7 commits

NQP bump brought: github.com/perl6/nqp/compare/2018....3-g8346970 8346970 [MoarVM Bump] Brings 2 commits 17a1eed Merge branch 'master' of github.com/perl6/nqp a065e99 rename and prettify with markdown ea1349b add note about speed-ups using NQP ... (7 more lines)
¦ rakudo: version bump brought these changes: github.com/perl6/nqp/compare/2018....3-g8346970
timotimo tbrowder: it may not be a good idea to suggest nqp usage to module authors that aren't at the same time core devs, since we make no stability guarantees for nqp:: ops at all 13:37
tbrowder how can that be since it is the basis of Rakudo Perl 6? 13:39
[Coke] timotimo++ 13:40
we can change nqp because we make rakudo still work when we do.
timotimo very easy: the core devs develop both rakudo and nqp, and when changes are made in nqp, the corresponding changes are made in rakudo immediately, then the version requirement is bumped
geekosaur tbrowder, because the higher level can provide guarantees
[Coke] so rakudo stays working despite the changes in nqp.
geekosaur if I write a line of C code, I can be reasonably certain it will compile to the same functionality on x86 or arm
which says nothing about the representation of the compiled code 13:41
timotimo some things are indeed very unlikely to change. atpos or existskey for example
but ops have been deprecated in the past, and it's not unlikely others will change behaviour in subtle ways that can trip up module authors that don't follow core development closely
tbrowder ok, so use of the stable rakudo is accompanied by its stable nqp, and an nqp user should have tests...
timotimo not only in stable rakudo 13:42
we increase the version requirement for nqp on the master branch multiple times in between releases 13:43
github.com/rakudo/rakudo/commits/m...P_REVISION - check out the history of this file
should we remove mention of the @rakudoperl twitter account from the nqp readme? its last tweet was a rakudo star release on 2013, the one before that was 2011 13:45
and its bio is "the perl 6 compiler for parrot" %)
wait, i saw that in the rakudo readme 13:47
it should probably link to p6lert, too
Geth nqp: ca77cf86e0 | (Zoffix Znet)++ (committed using GitHub Web editor) | README.pod
Correct note suggesting use of NQP

We don't offer any end-user support, so we shouldn't suggest people to use NQP for performance gains.
rakudo: 78ed66c704 | (Timo Paulssen)++ | README.md
remove mention of long-abandoned twitter account
rakudo: e09f2ecc41 | (Timo Paulssen)++ | README.md
point out the p6lert service near the beginning
Zoffix tbrowder: "how can it be": we can't remove or change behaviour of Perl 6's &lines() much, because people use that code and rely on its specific behaviour. If we change &lines, we'll break the users' code, because we have no guarantees we updated all of the code to use the new &lines correctly. However, we *can* change IO::Handle!LINES-ITERATOR any way we want and can even remove it, because it's a private 14:02
routine so the only use of it are inside Rakudo. We can update *all* of its usage while changing/removing it, thus no code gets broken. Same is with `nqp`, except unlike `!LINES-ITERATOR` we don't have an easy means to prevent other code using it, so we have to enforce it by documentation instead. Right now we can change `nqp::stat` to be more performant by modifying how it's called, but if users start
using NQP code, we can't easily change it without breaking code.
m: sub nqp-stat-s($f) { $f.IO.s }; sub file-size { nqp-stat-s $^f }; say file-size $*HOME.add: ".bashrc"
camelia 35
Zoffix m: sub nqp-stat-s($f, $stat) { $f.IO."$stat"() }; sub file-size { nqp-stat-s $^f, "s" }; say file-size $*HOME.add: ".bashrc" 14:03
camelia 35
Zoffix tbrowder: ^ above I changed how private nqp-stat-s is used and what it does, but the `file-size` routine remains unchanged and still gives the same results, even though under the hood it now uses a different NQP routine. Same with Rakudo. Yes, it requires stable NQP and NQP has tests, but it doesn't mean users are free to use NQP, because it change. 14:04
s/changed/removed nqp-stat entirely and added nqp-stat-s instead/;
tbrowder: and make enough people to use something unsupported and then this happens: irclog.perlgeek.de/perl6/2017-05-28#i_14650194 14:12
tbrowder ok, i get it, but authors of published modules are already using nqp (viz., drforr). 14:30
Zoffix tbrowder: that's not an argument to recommend more people do it. 14:32
AlexDaniel they're free to do that, and their code is free to break any time
Zoffix viz and DrForr will have their modules broken and will have only themselves to blame for using unsupported features.
AlexDaniel and obviously I will not be ensuring that breakage due to the use of nqp ops doesn't happen for the next release 14:33
tbrowder on that note, since cpan doesn’t show the current test status of modules, is there such a list somewhere else? i think zoffix runs tests periodically and lists some kind of results. 14:35
AlexDaniel you mean toaster?
this: toast.perl6.party/
tbrowder AlexDaniel: what about changing my comments to reflect the inhernt unstable nature of nqp?
AlexDaniel also I run toaster myself many times before the release, but I don't publish the results 14:36
tbrowder: Zoffix++ already fixed the note: github.com/perl6/nqp/commit/ca77cf...7108c00300
tbrowder ref toaster, i, as a module user, would like to see toaster results run regularly (i.e., automatically) and see results for linux and windows. 14:38
Zoffix: good mod, thnx 14:40
Zoffix tbrowder: well volunteered?
eco: Toaster
god's sake.
eco: Toaster 14:41
buggable Zoffix, Toaster 'Toaster of Perl 6 Ecosystem Modules': github.com/zoffixznet/perl6-Toaster
Zoffix tbrowder: you're free to donate the computation time and publish the results ^
tbrowder doesn’t perl6 have hosts to run such tests? 14:42
lizmat tbrowder: we don't have anything like cpan-testers (yet) 14:43
Zoffix tbrowder: we have a VM that could be used. You can speak to nine and he'd give you the keys to set up the software to make continuous Toaster run.
lizmat I hope the next PTS will allow us to set something up
Zoffix eco: Zef::CPANReporter
buggable Zoffix, Zef::CPANReporter 'Zef plugin to send test reports to cpantesters': github.com/garu/Zef-Reporter
Zoffix We do, in a way ^. Just need to find people to use it 14:44
tbrowder i’m happy to use my server while i’m still kicking if i can get som advice for setup..
Zoffix tbrowder: this should be it: github.com/zoffixznet/perl6-Toaste...e-vm-setup
tbrowder: but I found sometimes the process hangs, which makes automating it entirely a bit problematic
(more accurately, some module's installation hangs and the .kill that's supposed to kill them ain't working). 14:45
At least that happend in the past, but I think recently it was behaving OK. 14:46
tbrowder ok, looks like good starting instructions, but don’t hold your collective breaths... 14:55
Zoffix tbrowder: oh, and the entirely automated version would need to periodically nuke ~/.zef (or, possibly, just the cache dir inside). On my toaster VM in the past I did run out of disk space due to that dir being really big 15:05
tbrowder ok, i now have a dedicated server getting provisioned with Deb 8, 64-bit. anyone wanting access can send me their desired user name and a public key. i will give them sudoer privileges. ip tba when available. 15:51
oh, and the ip will be associated with domain perl6.club 15:55
Geth roast: ronaldxs++ created pull request #393:
Skip repeated socket close test on windows rather than hang.
roast: 8a4a0af471 | (Ronald Schmidt)++ | S32-io/IO-Socket-INET.t
Skip repeated socket close test on windows rather than hang.
roast: 86abb5376b | (Zoffix Znet)++ (committed using GitHub Web editor) | S32-io/IO-Socket-INET.t
Merge pull request #393 from ronaldxs/io-socket-inet-windows-hang

Skip repeated socket close test on windows rather than hang.
mr_ron Noticed that for WSL (windows subsystem linux) $*DISTRO.name is ubuntu but $*DISTRO.is-win is false. Wondering if $*DISTRO.is-win should be true on WSL? 18:51
Zoffix m: my class MuskRat2 is Rat is FatRat does Rational[Int, Int] {}; say MuskRat2.new(1, 2).perl; 19:17
camelia Cannot resolve caller infix:<==>(DeT, Int); none of these signatures match:
(\a, \b)
(Real \a, Real \b)
(Int:D \a, Int:D \b)
(int $a, int $b)
(Num:D \a, Num:D \b --> Bool)
(num $a, num $b --> Bool)
.oO( DeT, Int ? some weird bug that bakes in type captures? )
mr_ron: isn't WSL meant to be a ubuntu on windows and parts that are broken are just 'cause they're not implemented? I'd expect a .is-win to give me a windows filesystem and cmd.exe for a shell, not make it essentially meaningless. A lot of our tests operate on that belief. We could have an easier method to detect WSL, but shoving it into .is-win is a big mistake IMO 19:22
Don't have my Win10 box by my side, but isn't `$*KERNEL.release.contains: "Microsoft"` essentially it when it comes to WSL? 19:28
Doesn't look like a huge amount of typing, given you'd not need to detect it as often as differences between Win32 and linux 19:29
mr_ron r: dd $*KERNEL 19:57
camelia Kernel linux = Kernel.new(release => Str, name => "linux", auth => "unknown", version => Version, signature => Blob, desc => Str)
mr_ron Please forgive my ignorance but why isn't Kernel.release expanded? 20:01
r: say $*KERNEL.release
camelia ( no output )
#1 SMP Sun Oct 22 06:56:24 UTC 2017 (1d80e8a)
Zoffix What do you mean edxpanded? Do you mean why it's not part of `dd`'s output above? It's probably not an attribute 20:02
s: $*KERNEL, 'release', \()
SourceBaby Zoffix, Sauce is at github.com/rakudo/rakudo/blob/e09f...el.pm6#L60
Zoffix yeah just a method that shells out
mr_ron Got it thanks - will fix my patch to use $*KERNEL unless you object ... 20:03
Zoffix mr_ron: well, does it work? I didn't test it and was just going off this: github.com/MoarVM/MoarVM/commit/45...921d11R815 20:04
which'd probably be available in $*KERNEL.release
mr_ron My patch works but seems not to use best way to test for WSL. 20:05
Zoffix yeah 20:06
m: dd $*KERNEL.Capture 20:08
camelia \(:auth("unknown"), :desc(Str), :name("linux"), :release("#1 SMP Sun Oct 22 06:56:24 UTC 2017 (1d80e8a)"), :signature(Blob), :version(v4.4.92.31.default))
Geth roast: ronaldxs++ created pull request #394:
Fix wsl test thanks to explanation by Zoffix++. Oops
mr_ron Patch could still be better. Thinking cygwin or something could be another windows linux. Should really check if distro is ubuntu I think. 20:40
working on it now
never mind I guess - would expect other windows linux to have the same winsock delay problem. 20:43
dogbert17 m: 'fog'.subst: :g, /./, 'x'; say $/ 21:06
camelia 「g」
Geth roast: bbf13ab9c5 | (Ronald Schmidt)++ | S32-io/IO-Socket-INET.t
Fix wsl test thanks to explanation by Zoffix++. Oops
roast: 08b6dbeaf8 | (Zoffix Znet)++ (committed using GitHub Web editor) | S32-io/IO-Socket-INET.t
Merge pull request #394 from ronaldxs/io-socket-inet-windows-hang

Fix wsl test thanks to explanation by Zoffix++. Oops
dogbert17 Zoffix: it looks as if your commit github.com/rakudo/rakudo/commit/95...a3bd2db657 might have fixed RT #129912 :-) 21:26
synopsebot RT#129912 [new]: rt.perl.org/Ticket/Display.html?id=129912 [OPTIMIZER] Subset produces Cannot invoke this object (REPR: Null; VMNull)
MasterDuke timotimo: did you try benchmarking a bindpos_s version of the sql output? 21:47
timotimo i did not 21:50
dogbert17 Zoffix: shouldn't the test in github.com/perl6/roast/commit/7eea...aa0b0207ff be enough to close RT #126897 ? 22:47
synopsebot RT#126897 [open]: rt.perl.org/Ticket/Display.html?id=126897 [BUG] Slip objects deparse as List
Zoffix dogbert17: thanks. I closed those tickets. What ticket is `'fog'.subst: :g, /./, 'x'; say $/`? The fix should be trivial 23:05
dogbert17 Zoffix: let me check ... 23:06
RT #130688 23:08
synopsebot RT#130688 [new]: rt.perl.org/Ticket/Display.html?id=130688 [BUG] .subst/.subst-mutate with :g do not set $/ to all the matches
Zoffix Thanks 23:09
dogbert17 thanks for closing the other RT's I wasn't 100% certain hence the questions
Geth nqp: e98481898e | (Zoffix Znet)++ (committed using GitHub Web editor) | 2 files
Use correct extension for file format

Fixes github.com/rakudo/rakudo/issues/1597 R#1597
synopsebot R#1597 [open]: github.com/rakudo/rakudo/issues/1597 nqp/docs/jvminterop-goals.md is not in markdown format
Geth nqp: 50d436a7d0 | (Zoffix Znet)++ (committed using GitHub Web editor) | docs/jvminterop-goals.pod
Fix encoding error
dogbert17 Zoffix: slightly OT but perhaps this is something for you: www.vermintide.com/ 23:12
Zoffix I avoid Warhammer stuff. Doesn't strike me as something I'd enjoy 23:38