Geth star/fix-panda-jvm: 0abaabaf0f | (Zoffix Znet)++ | tools/build/Makefile.in
Swap panda to zef for JVM build
00:49
star: zoffixznet++ created pull request #109:
Swap panda to zef for JVM build
00:50
rakudo: ec5edcaec1 | (Zoffix Znet)++ | 2 files
Fix make test on Windows

It ain't got `strlen` in its standard library, so bake our own lib for the test and use that instead.
01:45
rakudo/ratlab-fattish-rat: c52a6fb3b2 | (Zoffix Znet)++ | 2 files
Crude trial impl of MidRat/MidRatStr
02:55
rakudo: 26522e8acd | (Zoffix Znet)++ | docs/archive/2018-03-04--Polishing-Rationals.md
Publish rev3 of "Polishing Rationals" proposal

  - Changes proposal of using RatStr as Rat/FatRat "allomorph"
   into `MidRat` type that is the actual Rat/FatRat allomorph
   (and `MidRatStr` being a `MidRat`/`Str` allomorph)
02:58
roast: ae92e139ab | usev6++ | S32-io/tell.t
Fix quoting for fudge message
06:34
star: 0abaabaf0f | (Zoffix Znet)++ | tools/build/Makefile.in
Swap panda to zef for JVM build
07:41
star: 95fc4c0376 | (Steve Mynott)++ (committed using GitHub Web editor) | tools/build/Makefile.in
Merge pull request #109 from rakudo/fix-panda-jvm

Swap panda to zef for JVM build
[Tux] Rakudo version 2018.02.1-136-g26522e8ac - MoarVM version 2018.02-33-ge639691a7
csv-ip5xs0.810 - 0.812
csv-ip5xs-207.650 - 7.740
csv-parser36.638 - 36.772
csv-test-xs-200.422 - 0.452
test9.070 - 9.108
test-t2.601 - 2.618
test-t --race1.077 - 1.081
test-t-2045.972 - 47.449
test-t-20 --race17.057 - 17.649
10:18
Geth star: c830bb063b | (Steve Mynott)++ | tools/build/Makefile.in
zef-m to zef-j, probably still doesn't work
12:15
rakudo/ratlab-role: 79bf17f53c | (Zoffix Znet)++ | 2 files
Stash mix-in-a-role approach as laternative to MidRat

Doesn't compile, but I see a problem with it already and giving up on it for now.
14:15
jnthn m: say Rat(+"1e-8") 14:20
camelia 0
jnthn m: say FatRat(+"1e-8")
camelia 0
jnthn Am I wrong to be surprsied that comes out as 0?
m: say (+"1e-8").Num 14:21
camelia 1e-08
jnthn m: say (+"1.5e-8").Num
camelia 1.5e-08
jnthn m: say (+"1.5e-8").Rat
camelia 0
dogbert2 Zoffix: speaking about Rats. It looks as if your commit github.com/rakudo/rakudo/commit/f6...9933000cb3 might have fixed RT #116423 14:24
synopsebot RT#116423 [new]: rt.perl.org/Ticket/Display.html?id=116423 [BUG] say 1.0000000000000000000000000000001 makes wrong denominator
Zoffix jnthn: sorta. The default epsilon in Num -> Rat is 1e-6 14:25
m: say (+"1.5e-8").Rat: 1e-8
camelia 0.000000015 14:26
Zoffix maybe it should be bumped to -15 or whatever the max mantissa in Num is
s: 0e0, 'Rat' 14:27
SourceBaby Zoffix, Sauce is at github.com/rakudo/rakudo/blob/2652...um.pm6#L46
Zoffix maybe not... "# However, to find that Rational would take more processing." 14:28
m: say 1.0000000000000000000000000000001.denominator.log: 2 14:30
camelia 102.979770941508
Zoffix dogbert17: not really, but that's what the "Polishing Rationals" work proposal will address. 14:31
dogbert2 aha
jnthn Zoffix: ah, thanks
dogbert2 m: say 1.0000000000000000000000000000001 14:32
camelia 1.0000000000000000000000000000001
Zoffix m: say 1.1111111111111111111111111111111111111111111111111
camelia 1.11111111111111110716365146799944341182708740234375
dogbert2 and it all came crashing down :-)
m: sub postfix:{}($a) { say "$a bracey brace" }; 42{} 14:34
camelia ===SORRY!===
This type cannot unbox to a native string: P6opaque, Block
Zoffix m: sub postfix:{}(|) {} 14:35
camelia ===SORRY!===
This type cannot unbox to a native string: P6opaque, Block
Zoffix Locally I get "Internal error: find_var_decl could not find $_"
dogbert2 Zoffix: the message you got is mentioned in RT #114554 14:37
synopsebot RT#114554 [open]: rt.perl.org/Ticket/Display.html?id=114554 [BUG] Definition of zero-length postfix operator wrongly allowed in Rakudo
Zoffix Looks like there's also a bug with colonpaired constants being parsed as categoricals for subs or something 14:53
m: constant x:<x>
camelia 5===SORRY!5=== Error while compiling <tmp>
Cannot add tokens of category 'x'
at <tmp>:1
------> 3constant x:<x>7⏏5<EOL>
Zoffix m: constant infix:<z> = *+*; say 4 z 2 14:54
camelia 5===SORRY!5=== Error while compiling <tmp>
Variable '&infix:<z>' is not declared
at <tmp>:1
------> 3constant infix:<z> = *+*; say 4 7⏏5z 2
Zoffix m: constant &infix:<z> = *+*; say 4 z 2
camelia 6
Zoffix m: BEGIN my &infix:<z> = *+*; say 4 z 2
camelia 6
Zoffix oh, I guess that's normal then
dogbert2 Zoffix: I'm rummaging through RT looking for issues which can be closed 14:56
Zoffix m: my $z:{:42foo} 14:59
camelia 5===SORRY!5=== Error while compiling <tmp>
You can't adverb $z
at <tmp>:1
------> 3my $z:{:42foo}7⏏5<EOL>
Zoffix m: my \z:{:42foo}
camelia ===SORRY!===
Cannot find method 'ast' on object of type NQPMu
Zoffix added as comment on 114554 15:01
dogbert2 jnthn: is stuff like this allowed? 'Thread.start({ say "entering"; require LWP::Simple; say "leaving" })' i.e. require in a thread? 15:17
jnthn dogbert2: Should be OK 15:18
Though hmm, one might race on global merging, but we can protect that if it ain't already
dogbert2 cool, then it might be possible to close, if tests are available, RT #126587 15:19
synopsebot RT#126587 [new]: rt.perl.org/Ticket/Display.html?id=126587 [SEGV] require inside thread segfault hang
dogbert2 because it seems to work and valgrind does not complain
Geth roast: 3222358fe8 | (Zoffix Znet)++ | S12-introspection/meta-class.t
Spec failed .^find_method/.^lookup

It returns nqp::null, which HLLized to Mu. While that differs from
  `Nil` we return in many other places to indicate "no value", IMO
it's OK to deviate from that convention in lower-level Metamodel code, as we can then rip the benefits of simpler testing of return value (e.g. nqp::isnull(nqp::callmethod(Str.HOW, :name<lookup>, Str, "bar"))) We also use nqp::null in a bunch of other Meta methods, like
  ::Versioning.ver and ::MethodContainer.cache_get
19:20
roast: a8a2cbd17e | (Zoffix Znet)++ | S12-introspection/meta-class.t
Revert "Spec failed .^find_method/.^lookup"

This reverts commit 3222358fe87c5cd1f66aaedef16ecba36d52dbb2.
Just spotted a description in the docs that I like better:
  "If no method can be found, it returns a VM-specific sentinel
value (typically a low-level NULL value)."
19:24
AlexDaniel releasable6: status 19:32
releasable6 AlexDaniel, Next release in 7 days and ≈23 hours. Blockers: github.com/rakudo/rakudo/issues?q=...%9A%A0%22. Changelog for this release was not started yet
Geth roast: 931e8d0e17 | (Zoffix Znet)++ | S12-introspection/meta-class.t
Vaguelly spec failed .^find_method/.^lookup

Spec that one can do a .defined test to check whether anything was found, but be vague about the actual value, to allow impls to use whatever value they get the most benefit from.
19:33
nqp: 7bc7ff9b0b | usev6++ | src/vm/jvm/runtime/org/perl6/nqp/io/StandardWriteHandle.java
[JVM] Keep track of bytes written to stdout/err

Makes test for RT #132254 pass.
19:43
synopsebot RT#132254 [resolved]: rt.perl.org/Ticket/Display.html?id=132254 [WIN32] $*OUT.tell always gives 0
dogbert17 m: 1 + Numeric.new # ho hum 20:06
camelia WARNINGS for <tmp>:
MoarVM panic: Memory allocation failed; could not allocate 83968 bytes
Geth roast: 1facfbf4b3 | usev6++ | S05-substitution/subst.t
Run test (todo) that doesn't die anymore
20:16
gfldex jnthn: was it a deliberate choice to have Channel.receive throw instead of returning a Failure? 22:55
timotimo someone could have sent a failure across the channel 22:56
gfldex Undefinedness is hard. Let's go shopping! :-> 22:59
jnthn gfldex: Yes 23:10
Note that for $channel.list { } and whenever $channel { } both handle termination "naturally" by ending the loop. 23:11