travis-ci Rakudo build failed. Zoffix Znet 'Bump NQP' 00:05
travis-ci.org/rakudo/rakudo/builds/285823953 github.com/rakudo/rakudo/compare/4...1f4ea26629
buggable [travis build above] ✓ All failures are due to: GitHub connectivity (1 failure).
MasterDuke any reason `method foo(\var) { if nqp::isconcrete(var) { bar() } else { baz() } }` couldn't be turned into `proto method foo(|) { * }; multi method foo(Mu:D \var) { bar(var) } ; multi method foo(Mu:U \var) { baz(var) }`? 02:28
[Tux] This is Rakudo version 2017.09-362-ge11f4ea26 built on MoarVM version 2017.09.1-582-gbd8e6687 06:09
csv-ip5xs 1.209 - 1.230
test 9.720 - 9.940
test-t 3.142 - 3.206
csv-parser 0.874 - 0.893
test-t 3.164 - 3.166 06:17
Zoffix MasterDuke: does it really need to be? 07:48
As in, the measurements show it using almost a meg more RAM with no speed benefit. 07:50
Hm, Perl's `system` also has this issue that `system 'perl6', '-e', 'say q|foo"bar|';` doesn't work 08:03
Which I guess means we have no problem and I should stop trying to find a solution \o/
And also means we should try to get a proper executable instead of a batch file on windows 08:05
lizmat Files=1227, Tests=75660, 346 wallclock secs (14.51 usr 5.00 sys + 2414.11 cusr 218.26 csys = 2651.88 CPU) 08:11
Zoffix bah, and this crap is exploitable too >:( 08:28
`run $*EXECUTABLE, '-e', '"&whoami';` on windows executes `whoami` on the shell 08:29
pretty serious bug come to think of it... Despite Perl's `system` crapping out on `system 'perl6', '-e', 'say q|foo"bar|';` so far I'm unable to make it execute stuff 08:35
lizmat is going to be afk for most of the day 08:46
Zoffix ah, k, `system 'C:\rakudo\perl6.bat', '-e', '"" &whoami';` does fall into shell 08:54
I guess I'll just pop open a [SECURITY] bug then 08:55
jnthn MasterDuke: I'd lean towards multis generally, though spesh is often smart enough to eliminate the branches when specializing, and so those two caess might well fall out the same post-optimization these days. 08:59
Smaller candidates are more likely to get inlined
Which is a big win
but given we also look at the post-speciallization size as well as the pre-specialization size, that may or may not matter 09:00
It depends. If bar/baz are themselves inlined into the caller then it's possible both the unoptimized and optimized would be over the limit
Of course, this is all stuff that we tweak in the VM too 09:01
Zoffix Filed rt.perl.org/Ticket/Display.html?id=132258 09:14
gah "Cannot delete C:\rakudo\install\bin/moar.dll: at -e line 1" 09:25
K, windows is positively unfun
huh, there were two moar.exe running still.... despite nothing on my terminal running 09:27
Geth rakudo/nom: 880a8e1c14 | (Zoffix Znet)++ | t/packages/Test/Helpers.pm6
Make is-run() mangle run() args on Windows

Due to RT#132258, when we `run $*EXECUTABLE` it goes throw cmd.exe and the tests fail due to bad quoting. Mangle the args to the test routine, so they get sent off properly.
The issue with sending double quotes appear to still remain in some ... (6 more lines)
10:07
synopsebot RT#132258 [new]: rt.perl.org/Ticket/Display.html?id=132258 [SECURITY][WIN32] `run "perl6" ...` can be made to execute shell commands
rakudo/nom: version bump brought these changes: github.com/perl6/nqp/compare/2017....5-ga65dbe0
47faae2b13 | (Zoffix Znet)++ | t/05-messages/01-errors.t

  - Use `is-run` routine for failing tests that does a bit of arg
   mangling to make them palatable to Windows's cmd.exe (RT#132258)
  - Skip one of the tests on Windows. Couldn't figure out how to send
   the quote in the args; the best I got resulted in duplicated quotes.
   Perhaps if there's a good solution to RT#132258 this fudge could be
   removed.
rakudo/nom: version bump brought these changes: github.com/perl6/nqp/compare/2017....5-ga65dbe0
b8eda66c90 | (Zoffix Znet)++ | tools/build/NQP_REVISION
¦ rakudo/nom: version bump brought these changes: github.com/perl6/nqp/compare/2017....5-ga65dbe0
Zoffix bah
thought it'd ask me if it were gonna do a merge commit. Oh well 10:08
Learning Perl 6 published September preview copy; for those who were wondering what's happening with it... 10:52
217 pages; with the final book aimed at 300 10:53
travis-ci Rakudo build passed. Zoffix Znet 'Merge branch 'nom' of github.com/rakudo/rakudo into nom' 11:03
travis-ci.org/rakudo/rakudo/builds/285977216 github.com/rakudo/rakudo/compare/e...eda66c90cd
stmuk 1www.amazon.co.uk/learning-perl-kee...149197768X 11:27
thats very expensive
of course I'll buy it :) 11:32
Zoffix says gonna be released on Nov 30, eh 11:39
yoleaux 11:05Z <piojo> Zoffix: thanks for your work on repl-mode. Extremely helpful for use in cygwin.
Zoffix huh
wonder why
stmuk I've found book publishing dates v unreliable in the past .. usually months later than first promised 11:40
the .fr claimed 5 days earlier 11:41
Zoffix stmuk: BTW, RT#132258 goes into the Windows proc issue you were inquiring about a few days ago. Turns out our current `run` only works right if cmd.exe is not involved, and if you're `run`ing perl6, then it is 11:42
synopsebot RT#132258 [new]: rt.perl.org/Ticket/Display.html?id=132258 [SECURITY][WIN32] `run "perl6" ...` can be made to execute shell commands
Zoffix maybe there's a way to fix up perl6 batch script dunno 11:43
Zoffix &
stmuk Zoffix: thanks I'll take a look 11:45
Zoffix m: (my @a)[1] = 42; @a.List.perl.say; @a.List[0].perl.say 12:28
camelia (Mu, 42)
Nil
Zoffix The whole hole business seems all over the map.
m: (my @a)[1] = 42; @a.Slip.perl.say; @a.Slip[0].perl.say
camelia slip(Any, 42)
Any
Zoffix This is part of RT#131783 but the fact that Mus show up in .perl but are actually Nils when you access by index is quite weird 12:31
synopsebot RT#131783 [open]: rt.perl.org/Ticket/Display.html?id=131783 [LTA] :delete holes in Arrays get turned to Mus when coercing to List or Slip
Zoffix m: (my @a)[1] = Mu; @a.List.gist.say; @a.List.perl.say; @a.List[0,1].say 12:34
camelia ((Mu) (Mu))
(Mu, Mu)
(Nil (Mu))
Zoffix gonna make .perl/.gist use a Nil there instead of a Mu 12:36
it's not working :( 12:46
m: (my @a = <a b c>)[1]:delete; use nqp; say @a.List.map: -> \v{ nqp::ifnull(v,Nil).gist } 12:47
camelia (a (Mu) c)
Zoffix Filed as Rt#132261 12:57
synopsebot RT#132261 [new]: rt.perl.org/Ticket/Display.html?id=132261 Unclear what a hole in a List is
Geth roast: 24d5fcb486 | (Zoffix Znet)++ | S32-array/delete.t
Test holes with .List and .Slip

RT#131783: rt.perl.org/Ticket/Display.html?id=131783
13:02
synopsebot RT#131783 [open]: rt.perl.org/Ticket/Display.html?id=131783 [LTA] :delete holes in Arrays get turned to Mus when coercing to List or Slip
Zoffix m: (my @a)[1] = Mu; my @b is default(42) = @a.List; say @b 13:04
camelia [(Mu) (Mu)]
Zoffix m: (my @a)[1] = Mu; my @b is default(42) = @a.List[*]; say @b
camelia [42 (Mu)]
Zoffix Now I see why no one did the TESTNEEDED tickets during SQUASHathon. Went through a dozen of them or so and none are clear what even needs to be tested and two of them aren't even fixed. 15:46
Zoffix gives up on this endeavour 15:47
ugexe Zoffix: ive been getting those zombie processes on windows as well 15:51
Zoffix weird 15:56
ugexe i wonder if the MVM_io_eventloop_cancel_work (such as inside MVM_proc_kill_async) can be interrupted in such a way that either handles don't get closed or the signal not getting sent at all (e.g. never reaches the exit_cb for uv_process_kill) 16:04
Zoffix ZOFLAG: t/spec/S17-channel/stress.t 16:27
:)
Took 244s before I killed it. Gonna switch to just running spectests for now 16:28
Geth roast: 1b0b55a873 | (Zoffix Znet)++ | S32-list/seq.t
Adjust broken Seq consumation tests

The tests are not part of 6.c-errata.
The tests passed for the wrong reason. `eqv` will actually return True if the LHS and RHS are the same object; the test was passing because the Seq was touched in the guts of cmp-ok.
Fix by using different consumed Seqs in the test.
16:38
rakudo/nom: 3684384db1 | (Zoffix Znet)++ | lib/Test.pm6
Make cmp-ok() take its args raw

  - Lets users use `=:=` op as comparator
  - Prevents interference with other ops that rely on address
   equivalence to do their thang, such as `eqv`
16:41
roast: 4bc442622d | (Zoffix Znet)++ | S24-testing/13-cmp-ok.t
Test cmp-ok() can be used with =:= op

Rakudo impl: github.com/rakudo/rakudo/commit/3684384db1
16:44
ugexe zombie processes seem to be coming from cntrl+c during the spectest? 16:46
zombie moar processes, and some zombie 'Windows Command Processor' processes
Zoffix I did do ctrl+c during spectest, so possible maybe
There's some test that hangs. I think INET sock one 16:47
ZOFVM: Files=1277, Tests=152649, 155 wallclock secs (21.43 usr 3.83 sys + 3351.40 cusr 200.18 csys = 3576.84 CPU)
ugexe all the moar processes were under load until cntrl+c, at which time 2 processes closed and the other 4 dropped to 0% cpu utilization 16:48
Geth rakudo/nom: 54507ac94e | (Zoffix Znet)++ | src/core/traits.pm
Fix `Mu` as is default() on attributes

Fixes RT#132082: rt.perl.org/Ticket/Display.html?id=132082
Routine args default to `Any`, so the multi for is default with Mu was never reached, instead going to the multi that cries about unknown traits.
16:50
synopsebot RT#132082 [new]: rt.perl.org/Ticket/Display.html?id=132082 [BUG] "is default(Mu)" parse failure when used on attribute
Geth roast: c537352009 | (Zoffix Znet)++ | S02-names/is_default.t
Undfudge and expand `is default(Mu)` test

RT#132082: rt.perl.org/Ticket/Display.html?id=132082 Rakudo fix: github.com/rakudo/rakudo/commit/54507ac94e
16:51
bartolin Zoffix: i saw your ticket about a holin in a List. i remember struggling with this about a year ago -- maybe the discussion in the following issue and the mentioned PR (both closed) are of interest: rt.perl.org/Ticket/Display.html?id=128320 17:08
Zoffix: the concrete problem was a NPE on the jvm backend, but if i remember correctly, the underlying issue was List.iterator in combination with a hole
Zoffix bartolin: don't know about JVM, but pretty sure nqp::ifnul in the iterator will fix the issue—that's what AT-POS does. However, I imagine having to do that on all the elements all the time will be a performance penalty that's really too steep to pay just to handle such a fringe usecase 17:20
Without considering perf, it makes sense to shove a Nil into holes, but... vOv 17:21
Maybe that's what the holes should be filled with, instead of Mus.
Like, right on deletion
(or expansion)
m: my @a is default(42) = <a b c>; @a[1] := Nil; say @a.List 17:23
camelia (a Nil c)
Zoffix But then we're stuck with this problem. Though Nil kinda makes more sense. Since we say it's our "lack of value" indicator
More sense than Mu I mean. Mu also sucks because some things don't like it. So a hole creates an issue that an Array/List don't fit some places 17:24
my @a = <a b c>; @a[1]:delete; sub ($a, $b, $c) {}(|@a.List)
m: my @a = <a b c>; @a[1]:delete; sub ($a, $b, $c) {}(|@a.List)
camelia Type check failed in binding to parameter '$b'; expected Any but got Mu (Mu)
in sub at <tmp> line 1
in block <unit> at <tmp> line 1
Zoffix m: my @a = <a b c>; @a[1]:delete; sub bar { say $^x }; @a.List.map: &bar 17:26
camelia a
Type check failed in binding to parameter '$x'; expected Any but got Mu (Mu)
in sub bar at <tmp> line 1
in block <unit> at <tmp> line 1
Zoffix etc.
m: my @a = <a b c>; @a[1]:delete; say [===] @a.List 17:27
camelia Cannot resolve caller infix:<===>(Str, Mu); none of these signatures match:
($?)
(\a, \b)
(Enumeration:D \a, Enumeration:D \b)
(Int:D \a, Int:D \b)
(int $a, int $b)
(Num:D \a, Num:D \b)
(num \a, num \b --> Bool)…
Zoffix But if they were Nils, these all would work
s: [], 'List', \()
SourceBaby Zoffix, Sauce is at github.com/rakudo/rakudo/blob/5450...ay.pm#L358
Zoffix tries sticking ifnul up in there 17:30
haw, it did the trick. Though there's this `:view` arg thats undocumented, unused (in core) and unspecced and if you use it you don't get the nils 17:42
timotimo is that a performance-related optimization thingie? 17:43
a "i swear i won't modify anything, so please don't copy"
Zoffix timotimo: yeah
.ask lizmat do you know if the `:view` can be removed? It doesn't appear to be used in core, undocumented, and unspecced. I'm making the method shove Nils into holes to resolve RT#132261 and user would still get Mus with `:view` arg. github.com/rakudo/rakudo/commit/e9...d32736c739 17:45
yoleaux Zoffix: I'll pass your message to lizmat.
synopsebot RT#132261 [new]: rt.perl.org/Ticket/Display.html?id=132261 Unclear what a hole in a List is 17:46
timotimo oh, it's no longer used?
Zoffix grep -FR 'List' src | grep 'view' shows me just the method definition, so I'm guessing no
Geth rakudo/nom: e135121914 | (Zoffix Znet)++ | src/core/Array.pm
Make Array.List fill holes with Nil

  - Fixes RT#132261: rt.perl.org/Ticket/Display.html?id=132261
  - List.AT-POS already gives a Nil for a hole, but .iterator does
   not do similar insertion. Doing so in Array.List lets us avoid
   paying for this feature in List.iterator
  - This behaviour is closer to what happens with .Slip or .AT-POS,
... (7 more lines)
17:58
synopsebot RT#132261 [new]: rt.perl.org/Ticket/Display.html?id=132261 Unclear what a hole in a List is
roast: 2257ab353f | (Zoffix Znet)++ | S32-array/delete.t
Spec Array.List fills holes with Nils

RT#132261: rt.perl.org/Ticket/Display.html?id=132261 Rakudo fix: github.com/rakudo/rakudo/commit/e135121914
17:59
Zoffix & 18:00
m: (my @a is default(42) = <a b c>)[1]:delete; @a.Slip.say 18:02
camelia (a 42 c)
Zoffix m: (my @a is default(42) = <a b c>)[1]:delete; @a.Slip.List.say
camelia (a (Mu) c)
Zoffix crap
Wonder if my fix fixes that. Kinda were expecting a 42 up there in the second version
No, don't think it fixes anything. dam 18:07
.in 4d fix (my @a is default(42) = <a b c>)[1]:delete; @a.Slip.List.say, bruh
yoleaux Zoffix: I'll remind you on 14 Oct 2017 18:07Z
Zoffix And currently it uses Slip.List that just transplants $!todo and $!reified. 18:08
m: my @a = <a b c>; my @list := @a.Slip.List; say @list; @a[1] = 42; say @list 18:09
camelia (a b c)
(a 42 c)
Zoffix Fun
m: my @a = <a b c>; my @list := @a.Slip.List; @list[1] = 42; say @list 18:10
camelia (a 42 c)
Zoffix Even more fun >:(
Actually, no the first one is more fun. 18:11
Zoffix &&
AlexDaniel- greppable6: hello 18:55
greppable6 AlexDaniel-, No! It wasn't me! It was the one-armed man! Backtrace: gist.github.com/bf4dc72c8d3e930225...2e6a500da5
AlexDaniel- so if it's from CPAN, how do I link to the code? 18:57
Zoffix mp6o/dist/Dist::Name:cpan
AlexDaniel- I guess a link to all-modules will do :S
Zoffix or that 18:58
Geth nqp: ebac9feeb6 | (Elizabeth Mattijsen)++ | 2 files
Make NQP's BUILDPLAN more like Perl6's BUILDPLAN
20:24
travis-ci NQP build failed. Elizabeth Mattijsen 'Make NQP's BUILDPLAN more like Perl6's BUILDPLAN' 20:41
travis-ci.org/perl6/nqp/builds/286217065 github.com/perl6/nqp/compare/a65db...ac9feeb696
lizmat doesn't know if she just broke the JVM build or not 20:42
java.lang.ClassCastException: org.perl6.nqp.sixmodel.reprs.P6strInstance cannot be cast to java.lang.String
does that ring a bell?
bartolin no, that's a recently added test that fails on jvm 20:43
lizmat so it's not something I just broke in my last commit ? 20:44
lizmat is pretty noob in nqp proper
bartolin i'm pretty sure, i've seen that one before.
it uses dies-ok and it looks like there is something wrong with handling (native) java exception 20:45
pmurias bartolin: it's the same problem we had with the divide by zero 20:54
bartolin: the say call after the try block in dies-ok seems to cause problems 20:55
s/cause problems/works incorrectly because the try screws up something 20:56
lizmat ok 21:04
cool *phew*
AlexDaniel- releasable6: next 21:07
releasable6 AlexDaniel-, Next release in 10 days and ≈21 hours. No blockers. Changelog for this release was not started yet
AlexDaniel-, Details: gist.github.com/1b1b91f6ee10fd8322...21f9ddaf98
bartolin pmurias: yeah, i tried to understand what happens, but failed. maybe the exception isn't properly handled by CATCH -- and then blows up when 'ok' is called? 21:10
Zoffix Yeah, that nqp error existed for awhile 21:19
Geth rakudo/nom: dd943eded8 | (Elizabeth Mattijsen)++ | 4 files
Fix for RT #132236

  - Bump NQP
   - this brings the NQP buildplan in line with Perl 6
   - 0 -> CODE, 1 -> 0, 2 -> 11, 3 -> 12
  - Add support for task codes 10, 11, 12 to auto-generated BUILDALL
   - having task 10 makes it ready for auto-generated BUILD_LEAST_DERIVED
  - Add support for tasks 11,12 to Mu.BUILDALL and Mu.BUILD_LEAST_DERIVED
   - in case somehow we failed to auto-generate a BUILDALL
21:27
synopsebot RT#132236 [open]: rt.perl.org/Ticket/Display.html?id=132236 [REGRESSION] Meta object construction
lizmat calls it a day again 21:32
Geth rakudo/nom: 3ca6554fdd | (David Warring)++ | 3 files
implement typed pointer increment and array dereference

This enables:
   use NativeCall;
   my CArray[uint16] $a .= new: 10, 20 ... 100;
   my $p = nativecast(Pointer[uint16], $a);
... (6 more lines)
22:04
rakudo/nom: e19b3409e8 | (Zoffix Znet)++ (committed using GitHub Web editor) | 3 files
Merge pull request #1186 from dwarring/pointer++

implement typed pointer increment and array dereference salortiz++
Zoffix dam wtf... placed `note($l.HOW.name($l)) if nqp::atkey(nqp::getenvhash(), 'ZZZ');` above this line github.com/rakudo/rakudo/blob/nom/...ar.nqp#L68 trying to dump the value, but when I set the env key make it dump, I get "Cannot invoke this object (REPR: Null; VMNull)" :/ 22:06
nqp: note(42.HOW.name(42)) if nqp::atkey(nqp::getenvhash(), 'ME')
camelia BOOTInt
Zoffix And here it works fine :\
k, note() nor say() ain't working in that place. nqp::say() worked 22:16
timotimo yo zoffix, whatcha up to? 22:17
Zoffix m: $ = ""; sub postfix:<♥> { $^a + 42}; say "{ 5♥ }" 22:18
camelia 5===SORRY!5=== Error while compiling <tmp>
Bogus postfix
at <tmp>:1
------> 3"; sub postfix:<♥> { $^a + 42}; say "{ 57⏏5♥ }"
expecting any of:
infix
infix stopper
statement end
statement m…
Zoffix timotimo: debugging this. Well, I already know it's %quote_lang_cache that's the cause, but just trying to figure out why the Slang Refactor bustred it
timotimo ah, that 22:19
yeah, man, that's a wild bug
Zoffix oh lol, I see it I think 22:25
oh, false alarm, was looking at the wrong place -_- 22:26
timotimo boy do i know that feeling
jnthn Zoffix++ # persistence 22:27
I don't have any guesses for you, alas
Zoffix :) 22:29
MasterDuke jnthn, Zoffix: thanks for the answers. i'm trying to make INTERPOLATE a little faster, so i'd only make the change if it does measure better, but i wanted to make sure the two options i asked about were in fact logically the same 22:31
Zoffix MasterDuke: FWIW, I was measuring with /usr/bin/time perl6 -e 'use nqp; class Foo { method foo (\var) { nqp::isconcrete(var) ?? rand !! rand } }; my $f = Foo; $ = Foo.new.foo: 42 for ^3000_000; say now - INIT now' 22:33
vs same but with multies
MasterDuke i believe one of the branches is taken way more than the other, so trying to break it out into its own more self-contained multi 22:35
Zoffix cool 22:36
MasterDuke but you saw a 1m more ram for the multi version? wouldn't have expected that
Zoffix buggable: zen I WILL EAT THIS BUG ALIVE!!! 22:37
buggable Zoffix, "When the mind is perfectly clear, what is is what we want."
AlexDaniel- greppable6: Bundesliga 22:40
greppable6 AlexDaniel-, gist.github.com/1100fce50f199befe1...c54dcc2e2c
AlexDaniel- greppable6: HandleSupplier
greppable6 AlexDaniel-, gist.github.com/188793862e774b9ab1...5d8b910336
AlexDaniel- gitlab and cpan modules are now supported 22:41
Zoffix AlexDaniel-++
AlexDaniel-
.oO( -++ :D )
MasterDuke IIRC, protos currently don't support code in their body? that is actually shared by all the multis? 22:45
Zoffix MasterDuke: they do. You're thinking of regex protos 22:46
m: proto x { say 'meow'; {*}; say 'moo' }; multi x(Int) { say 'here' }; multi x(Num) { say 'there' }; x 42; x pi
camelia Too many positionals passed; expected 0 arguments but got 1
in sub x at <tmp> line 1
in block <unit> at <tmp> line 1
Zoffix m: proto($) x { say 'meow'; {*}; say 'moo' }; multi x(Int) { say 'here' }; multi x(Num) { say 'there' }; x 42; x pi
camelia 5===SORRY!5=== Error while compiling <tmp>
{*} may only appear in proto
at <tmp>:1
------> 3proto($) x { say 'meow'; {*}7⏏5; say 'moo' }; multi x(Int) { say 'here'
expecting any of:
horizontal whitespace
term
Zoffix it is a proto, bruh 22:47
m: proto sub($) x { say 'meow'; {*}; say 'moo' }; multi x(Int) { say 'here' }; multi x(Num) { say 'there' }; x 42; x pi
camelia 5===SORRY!5=== Error while compiling <tmp>
Missing block
at <tmp>:1
------> 3proto sub($)7⏏5 x { say 'meow'; {*}; say 'moo' }; multi
expecting any of:
new name to be defined
Zoffix buggable: zen KILL IT WITH FIRE!
buggable Zoffix, "Today, you can decide to walk in freedom. You can choose to walk differently. You can walk as a free person, enjoying every step."
jnthn Zoffix: Why are you putting the signature before the name? ;)
Zoffix Ah
m: proto x($) { say 'meow'; {*}; say 'moo' }; multi x(Int) { say 'here' }; multi x(Num) { say 'there' }; x 42; x pi
camelia meow
here
moo
meow
there
moo
ugexe hmm, mkdir("exists.txt") returns "exists.txt".IO 22:56
e.g. mkdir on file
timotimo that probably shouldn't happen 23:00
AlexDaniel- “hmm” indeed. mkdir -p complains about this kind of stuff
MasterDuke Zoffix: ah, thanks 23:01
Zoffix ugexe: yeah, there's a ticket for that: rt.perl.org/Ticket/Display.html?id...et-history 23:07
MasterDuke making progress, now INTERPOLATE is 3x slower 23:21
that's what people want, right? 23:22
Zoffix :) 23:23
1. Make it slower for a few releases
2. Fix it back to past speed
3. Log "Made INTERPOALTE 3x faster" :) 23:24
???
PROFIT!
AlexDaniel- 1½. Log “Made INTERPOALTE 3x slower”
Geth nqp: 26abb152bb | (Zoffix Znet)++ | tools/build/MOAR_REVISION
Bump MoarVM
23:25
¦ nqp: version bump brought these changes: github.com/MoarVM/MoarVM/compare/2...1-gf2deedf
rakudo/nom: a72214c4f1 | (Zoffix Znet)++ | tools/build/NQP_REVISION
Bump NQP
¦ rakudo/nom: version bump brought these changes: github.com/perl6/nqp/compare/2017....7-g26abb15
MasterDuke heh. rather surprised my changes had this much effect, thought they were pretty minot so far 23:26
*minor
Zoffix :S even with the brute-force hammer of nuking the cache, the bug is still there. The output says clearing cache-blah blah but shows zero keys in cache clearing method (both before and after deletion) yet in quote_lang method the keys still show up :/ 23:34
gist.github.com/zoffixznet/ff4aa31...2d9d44ead2
Gonna continue it tomorrow
travis-ci NQP build failed. Zoffix Znet 'Bump MoarVM' 23:38
travis-ci.org/perl6/nqp/builds/286282233 github.com/perl6/nqp/compare/ebac9...abb152bbf3
MasterDuke hm, after my changes tons of time is spent in find_best_dispatchee. it doesn't like something about my multis 23:46