buggable ๐ŸŽบ๐ŸŽบ๐ŸŽบ It's time for the monthly Accidental /win Lottery ๐Ÿ˜๐Ÿ˜๐Ÿ˜ We have 10 ballots submitted by 7 users! DRUM ROLL PLEASE!... 00:00
And the winning number is 23! Congratulations to raschipi! You win a roll of duck tape!
๐ŸŽบ๐ŸŽบ๐ŸŽบ It's time for the monthly Accidental /win Lottery ๐Ÿ˜๐Ÿ˜๐Ÿ˜ We have 0 ballots submitted by 0 users! DRUM ROLL PLEASE!...
And the winning number is 42! Congratulations to Zoffix! You win a can of WD40!
๐ŸŽบ๐ŸŽบ๐ŸŽบ It's time for the monthly Accidental /win Lottery ๐Ÿ˜๐Ÿ˜๐Ÿ˜ We have 0 ballots submitted by 0 users! DRUM ROLL PLEASE!...
And the winning number is 42! Congratulations to Zoffix! You win a roll of duck tape!
๐ŸŽบ๐ŸŽบ๐ŸŽบ It's time for the monthly Accidental /win Lottery ๐Ÿ˜๐Ÿ˜๐Ÿ˜ We have 0 ballots submitted by 0 users! DRUM ROLL PLEASE!...
And the winning number is 42! Congratulations to Zoffix! You win a roll of duck tape!
๐ŸŽบ๐ŸŽบ๐ŸŽบ It's time for the monthly Accidental /win Lottery ๐Ÿ˜๐Ÿ˜๐Ÿ˜ We have 0 ballots submitted by 0 users! DRUM ROLL PLEASE!...
And the winning number is 42! Congratulations to Zoffix! You win a can of WD40!
๐ŸŽบ๐ŸŽบ๐ŸŽบ It's time for the monthly Accidental /win Lottery ๐Ÿ˜๐Ÿ˜๐Ÿ˜ We have 0 ballots submitted by 0 users! DRUM ROLL PLEASE!...
And the winning number is 42! Congratulations to Zoffix! You win a can of WD40!
๐ŸŽบ๐ŸŽบ๐ŸŽบ It's time for the monthly Accidental /win Lottery ๐Ÿ˜๐Ÿ˜๐Ÿ˜ We have 0 ballots submitted by 0 users! DRUM ROLL PLEASE!...
And the winning number is 42! Congratulations to Zoffix! You win a roll of duck tape!
๐ŸŽบ๐ŸŽบ๐ŸŽบ It's time for the monthly Accidental /win Lottery ๐Ÿ˜๐Ÿ˜๐Ÿ˜ We have 0 ballots submitted by 0 users! DRUM ROLL PLEASE!...
And the winning number is 42! Congratulations to Zoffix! You win a roll of duck tape!
Zoffix wonder what happened with it :/ 00:01
buggable And the winning number is 42! Congratulations to Zoffix! You win a roll of duck tape! 00:03
๐ŸŽบ๐ŸŽบ๐ŸŽบ It's time for the monthly Accidental /win Lottery ๐Ÿ˜๐Ÿ˜๐Ÿ˜ We have 0 ballots submitted by 0 users! DRUM ROLL PLEASE!...
Zoffix :|
I'd wager latest rakudo got some bug with Promise.in or DateTime :/ 00:04
well, 2018.02.1 00:05
buggable And the winning number is 42! Congratulations to Zoffix! You win a roll of duck tape!
๐ŸŽบ๐ŸŽบ๐ŸŽบ It's time for the monthly Accidental /win Lottery ๐Ÿ˜๐Ÿ˜๐Ÿ˜ We have 0 ballots submitted by 0 users! DRUM ROLL PLEASE!...
And the winning number is 42! Congratulations to Zoffix! You win a roll of duck tape! 00:06
Zoffix Ah, nm, it's actually a bug fix to Promise.in with negative numbers hanging that was masking a bug I made with date calculation 00:11
m: say DateTime.now 00:13
camelia 2018-04-01T02:13:53.402277+02:00
Zoffix ^ that's still 31st on my server 00:14
m: say DateTime.now.later(:month).clone(:0timezone)
camelia 2018-05-01T02:14:16.996120Z
Zoffix which makes ^ that 1st of April, instantly triggering the thing :)
c: 2017.07,HEAD await Promise.in(-2).then: { say "meows" }
committable6 Zoffix, ยฆ2017.07: ยซยซtimed out after 10 secondsยป ยซexit signal = SIGHUP (1)ยปยป ยฆHEAD(dc1a3f5): ยซmeowsโคยป 00:15
AlexDaniel reportable6: 2018-03-01T00:00:00Z 2018-04-01T00:00:00Z 02:40
reportable6 AlexDaniel, OK, working on it! This may take up to 40 seconds
AlexDaniel, gist.github.com/45a2df11156178a841...d77296f174 02:41
Zoffix April fools! 07:07
:)
Geth_ nqp/nqpmake: 5 commits pushed by (Stefan Seifert)++ 09:24
nine No April fools joke - a make in NQP supporting parallel build in 355 LOC 09:25
Still got a few issues to solve, but all in all it's pretty much there 09:26
moritz nine: can it build NQP? :-) 09:49
Geth_ nqp/nqpmake: bd279a021f | (Stefan Seifert)++ | tools/make.nqp
Fix error if less than 3 jobs are buildable initially
09:55
nqp/nqpmake: cfff514fc3 | (Stefan Seifert)++ | tools/make.nqp
Fix error in detection of prerequisite updates
nqp/nqpmake: 800a6ef3c1 | (Stefan Seifert)++ | tools/make.nqp
Fix hang when we don't need to build a target
nine This takes care of the issues I mentioned.
moritz: it can indeed
moritz woah 09:56
nine nqp and rakudo but not yet MoarVM as the latter uses GNU extensions in its Makefile and I've yet to add support for those 09:57
Geth_ nqp/nqpmake: 20d1249bda | (Stefan Seifert)++ | tools/make.nqp
Build first non-special target if no explicit target is given
10:03
robertle_ nine: is it free of the weird default rules? that would make it a better make than the real one (in my opinion at least) 10:19
implicit rules, that is 10:20
Kaiepi i think the new nqp makefile looks a lot more legible 11:01
timotimo did the makefile itself change, too? 11:02
Kaiepi nah, this looks like it parses and runs them 11:04
nine robertle_: I implemented exactly as much as rakudo and nqp need, so no default rules yet 11:15
Kaiepi can't wait ofor rakudo to get faster to compile 11:20
timotimo oh the plight of the core developer :) 11:24
Kaiepi i've noticed that :3
nowhere near as bad as llvm but irksome when debugging something minor 11:25
timotimo you can often get around that by prototyping your changes in a smaller file together with tests, if needed you can use "augment" to change existing classes 11:27
Kaiepi i've been doing that a little bit of that for changes to rakudo 11:30
lots of test.c, test.o, libtest.so floating around from nativecall tinkering
speaking of which 11:31
would a sub to preload symbols from libraries be a good idea to have for other libraries that need them to run? 11:32
timotimo how do you mean? 11:33
if you mean nativecall, if the libraries are "easy to find", then the dynamic linker would - i assume - find them by themselves when they are needed 11:34
Kaiepi editline needs symbols from ncurses, termcap, and wherever FILE comes from, but i don't have a very good way to load them 11:35
best i can do for now is run some of their subs in an INIT phaser in each module editline's in 11:36
same thing happens to readline on freebsd as well 11:37
timotimo isn't FILE just a #define or typedef?
Kaiepi i'd need to check, but i think it differs between systems 11:38
timotimo of course it does %) 11:41
it's not very fun to have to deal with #define things with an FFI 11:42
because the only way to get at them (if the library doesn't cooperate) is to parse the header files
Kaiepi yikes 11:43
timotimo well, another way is to ask a C compiler to do that for you 11:44
buggable: eco gptrixie
buggable timotimo, App::GPTrixie 'Generate NativeCall code from C headers file': github.com/Skarsnik/gptrixie
Kaiepi that looks really useful 11:50
timotimo it's not easy to set up, though. you need the exact right version of this castxml thing 11:51
dogbert17 timotimo: I bleive there might be a bug in your 'dont_gc_in_spesh' branch 11:53
*believe
timotimo cool
dogbert17 several small scripts makes valgrind complain 11:54
==13920== Conditional jump or move depends on uninitialised value(s)
==13920== at 0x50AAE83: worker (worker.c:16)
I guess that's here: github.com/MoarVM/MoarVM/blob/mast...h/worker.c
does it seem plausible? 11:55
timotimo: do you need more info, e.g. example code etc 12:01
timotimo huh? there is no conditional there though? 12:02
dogbert17 the next line in the trace is github.com/MoarVM/MoarVM/blob/mast...eads.c#L59 12:04
timotimo yeah, that's how the worker gets launched 12:05
that example code could be interesting 12:06
dogbert17 ok, sec 12:07
gist.github.com/dogbert17/8156ed11...40f81ee24e 12:08
timotimo does it only happen a single time? 12:12
dogbert17 yeah
timotimo only in my branch?
dogbert17 it doesn't happen on HEAD 12:13
timotimo pretty weird.
dogbert17 yes 12:14
timotimo wait, line 16
of course, that's an MVMROOT 12:15
everything inside the { } will be treated as "on line 16"
can you throw out the MVMROOT2 line and the corresponding }); and instead put:
dogbert17 the ROOT of all problems :)
timotimo MVM_gc_temp_root_push(tc, (MVMCollectable**)&updated_static_frames); MVM_gc_temp_root_push(tc, (MVMCollectable**)&previous_static_frames);
and at the bottom MVM_gc_temp_root_pop_n(tc, 2) 12:16
;
dogbert17 hmm, let me try 12:17
tbrowder_ .tell Zoffix i took a quick look at doc build and saw a couple msgs about pod table issues but i canโ€™t duplicate it again
yoleaux tbrowder_: I'll pass your message to Zoffix.
dogbert17 timotimo: src/spesh/worker.c:16:5: warning: implicit declaration of function โ€˜MVM_gc_temp_root_pushโ€™ [-Wimplicit-function-declaration] 12:18
timotimo oh, switch root and temp around
dogbert17 ./libmoar.so: undefined reference to `MVM_gc_temp_root_push' 12:20
I had forgotten a semicolon
timotimo yeah, has to be root_temp not temp_root 12:21
dogbert17 ah
ok, now it builds 12:22
now I get: ==20304== Conditional jump or move depends on uninitialised value(s) 12:23
==20304== at 0x50AAE83: worker (worker.c:45)
so a different line
the 'MVM_spesh_candidate_add' line 12:24
eh, I'm full of it :( it's another ' MVMROOT(tc, sl, {' 12:25
timotimo :D 12:28
dogbert17 so something fishy is going one here 12:34
lizmat Files=1238, Tests=76257, 307 wallclock secs (14.46 usr 5.24 sys + 2104.57 cusr 201.37 csys = 2325.64 CPU) 12:59
Geth_ nqp: 6b49767ec5 | usev6++ | 2 files
[JVM] Fix nqp::takedispatcher in non-void context

Fixes github.com/perl6/nqp/issues/384
13:11
nqp: 32fd547c10 | (Jeremy Studer)++ | src/vm/moar/profiler/template.html
Profiler Template: Allocations modal close by btn

Make Allocations modal closable by the corner X and via 'Close' button in addition to clicking outside. Small usability improvement.
13:51
[Tux] Rakudo version 2018.03-137-gdc1a3f5a3 - MoarVM version 2018.03-59-gfc1bb3126
csv-ip5xs0.911 - 0.927
csv-ip5xs-208.984 - 9.152
csv-parser38.444 - 41.819
csv-test-xs-200.451 - 0.465
test9.849 - 10.822
test-t2.555 - 2.583
test-t --race1.139 - 1.188
test-t-2045.505 - 46.150
test-t-20 --race17.760 - 18.243
15:14
dogbert17 anyone seen this before: 'This type (<anon|34>) does not support positional operations' 15:25
it shows from time to time when running: ./perl6 --profile --profile-filename=race1.json -e 'say (^400).race(batch=>1, degree=>4).map({$_}).grep(* > 12345).elems' 15:27
the example coming from RT #128280
synopsebot RT#128280 [open]: rt.perl.org/Ticket/Display.html?id=128280 [SEGV] core dump while profiling with a '.race'
jstuder dogbert17: I've seen that error msg on the P6Opaque REPR (github.com/MoarVM/MoarVM/blob/mast...e.c#L1225) 15:42
it looks like from time to time something calls low-level positional ops on an iterator 15:43
dogbert17 jstuder: I think you're right, for one of the crashes I got a backtrace and on top of it was: at SETTING::src/core/Iterator.pm6:69 (./CORE.setting.moarvm:push-until-lazy) 15:45
jstuder yeah, don't understand why it would be performing that op on the iterator itself though 15:51
dogbert17 when it fails the call which displays the error msg comes from here: github.com/MoarVM/MoarVM/blob/mast...ue.c#L1263 16:10
timotimo have you tried passing --ll-exception to perl6? 16:12
dogbert17 timotimo: no but once I got this: gist.github.com/dogbert17/d7a1a819...c425f7edab 16:14
timotimo yeah that's one of the iterator objects surely
dogbert17 gdb shows: #1 0x00007ffff7641e7f in push (tc=0x32fbe80, st=0x19e1900, root=0x333f5f8, data=0x333f610, value=..., kind=4) at src/6model/reprs/P6opaque.c:1263 16:15
timotimo and MVM_dump_backtrace?
dogbert17 let's see :)
exactly like the gist 16:16
timotimo mhm
well, the debugserver would help here :P
i don't have time right now
dogbert17 perhaps later 16:17
I've not had the chance to use that new cool feature yet :) 16:18
Kaypie .magikjokku 17:13
Auuuhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhino 17:14
timotimo accidentally your password? 17:15
lizmat Kaypie: van mij wel hoor :-) 17:44
pmurias should nqp::eqaddr(Str:U, $Str-U-from-a-different-module) be 1 everywhere? 17:48
it seems to change depending if the module I'm using it uses Hash[Str:U]
timotimo we probably should cache those 17:57
but i imagine unless you explicitly import the one from there you'll get a "custom" one? 17:58
pmurias timotimo: isn't that supposed to be cached by paramaterized type interning?
timotimo it'd probably only cache Hash[Str:U] if the Str:U exactly matches? 17:59
lizmat
.oO( a password starting with "/" is safer in the IRC world :-)
18:06
pmurias timotimo: if Hash[Str:U] is cached Str:U gets cached too 18:27
dogbert17 what's the correct way to install App::MoarVM::Debug? zef install doesn't seem to do it 18:31
during the test phase I get several 'last without loop construct' and then it fails 18:33
to clarify, the error messages comes when testing the dependency MoarVM::Remote:ver<0.0.1> 18:35
timotimo oh, the tests may need some very special setup 18:43
so try with --/test
dogbert17 will do, thx 18:44
Kaiepi timotimo lizmat my keyboard's on the fritz 19:47
but the aaauuuuhhhhhhhhh it typed out on its own is how i feel about it 19:48
Geth_ rakudo: 32b3ddc7e1 | (Elizabeth Mattijsen)++ | src/core/Order.pm6
Micro-optimize ORDER() by about 7%

This should help a lot of infix:<cmp> usage.
21:41
rakudo: jstuder-gh++ created pull request #1680:
Rewrite List cmp in NQP for faster execution
22:34
rakudo/master: 4 commits pushed by (Jeremy Studer)++, (Elizabeth Mattijsen)++ 23:29
rakudo: 111f67a4bc | (Elizabeth Mattijsen)++ | src/core/Promise.pm6
Micro-optimize Promise.vow

We can release the lock as soon as the Promise is marked that it has a vow. We don't actually need to create the vow before releasing the lock. So we don't need a variable to store the vow in.
23:43
lizmat and that concludes my hacking for today& 23:44