Perl 6 language and compiler development | Logs at colabti.org/irclogger/irclogger_log/perl6-dev | For toolchain/installation stuff see #perl6-toolchain | For MoarVM see #moarvm
Set by Zoffix on 27 July 2018.
01:31 ZzZombo left 02:54 ggoebel left 03:08 ggoebel joined 03:09 cog left, p6bannerbot sets mode: +v ggoebel 03:13 lizmat left 03:29 Ven` joined 03:30 p6bannerbot sets mode: +v Ven` 03:34 Ven` left 04:49 epony left 05:25 robertle left 06:20 ZzZombo joined, p6bannerbot sets mode: +v ZzZombo, ZzZombo left 06:21 ZzZombo joined, p6bannerbot sets mode: +v ZzZombo 06:22 robertle joined, p6bannerbot sets mode: +v robertle 06:27 patrickb joined, p6bannerbot sets mode: +v patrickb 06:47 epony joined, p6bannerbot sets mode: +v epony 06:56 brrt joined 06:57 p6bannerbot sets mode: +v brrt
releasable6 Next release in ≈2 days and ≈11 hours. 1 blocker. Please log your changes in the ChangeLog: github.com/rakudo/rakudo/wiki/ChangeLog-Draft 07:00
brrt what's the blocker? 07:04
07:47 lizmat joined 07:48 p6bannerbot sets mode: +v lizmat
lizmat timotimo: cool! 07:49
07:51 Ven` joined 07:52 p6bannerbot sets mode: +v Ven` 08:06 brrt left
|Tux| Rakudo version 2018.08-97-g0a416dadb - MoarVM version 2018.08-70-g1d68871d0
csv-ip5xs0.934 - 0.963
csv-ip5xs-207.325 - 7.596
csv-parser22.191 - 22.663
csv-test-xs-200.427 - 0.435
test9.277 - 9.354
test-t2.056 - 2.057
test-t --race0.855 - 0.936
test-t-2035.183 - 35.813
test-t-20 --race11.036 - 11.714
 08:21
08:34 brrt joined, p6bannerbot sets mode: +v brrt 08:35 Ven` left 08:43 x[LGWs4x4i]uG2N0 joined 08:44 p6bannerbot sets mode: +v x[LGWs4x4i]uG2N0
brrt robertle: use MoarVM tools/jit-bisect.pl --spesh 08:50
I have a plan to make that more automagic
08:50 Ven` joined 08:51 p6bannerbot sets mode: +v Ven`
AlexDaniel brrt: blocker: github.com/MoarVM/MoarVM/issues/951 08:51
releasable6: status
releasable6 AlexDaniel, Next release in ≈2 days and ≈10 hours. 1 blocker. 1 out of 97 commits logged (⚠ 46 warnings)
AlexDaniel, Details: gist.github.com/93a74f8df4c245b604...7b5d7ac496
brrt if it's a raspberry pi is, then I'm afraid the JIT isn't going to help much 08:52
well, the JIT isn't to blame, is what I mean
timotimo no JITs here, sir
09:32 brrt left 09:51 cognominal-p6 joined 09:52 p6bannerbot sets mode: +v cognominal-p6 09:53 x[LGWs4x4i]uG2N0 left 09:57 x[LGWs4x4i]uG2N0 joined 09:58 p6bannerbot sets mode: +v x[LGWs4x4i]uG2N0
lizmat Files=1251, Tests=76116, 342 wallclock secs (15.91 usr 5.65 sys + 2402.97 cusr 230.35 csys = 2654.88 CPU) 10:02
10:16 ZzZombo left 10:56 lizmat left 11:05 lizmat joined 11:06 p6bannerbot sets mode: +v lizmat 11:07 Ven` left
tbrowder__ hi, perl6-dev 11:17
a few days ago i asked about some grammar syntax (**1 and **0..1 quantifiers on a token reference see colabti.org/irclogger/irclogger_log...9-05#l98). 11:18
in this gist i show the uses and what i would like to try to change to a more modern syntax. comments welcome; gist: gist.github.com/tbrowder/ad7019a21...0f42ffb509 11:21
please use a dot command as i will be afk& 11:22
11:23 japhb_ left
AlexDaniel .tell tbrowder just try it? If it works then no problem (I'd expect some tests to fail if it makes a difference) 11:41
yoleaux AlexDaniel: I'll pass your message to tbrowder.
tbrowder__ AlexDaniel: I do plan to try, just wanting to see if there is some other choice to try 11:44
11:44 ZzZombo joined, p6bannerbot sets mode: +v ZzZombo 11:55 brrt joined 11:56 p6bannerbot sets mode: +v brrt, cog joined 11:57 p6bannerbot sets mode: +v cog 11:59 cognominal-p6 left 12:05 Ven` joined
Geth nqp: a7351c4a05 | (Timo Paulssen)++ | src/vm/moar/HLL/Backend.nqp
put the name of the SC in every type's extra info

this lets us figure out which types come from the raukdo core setting, the bootstrap, or nqp ... or the user's code, or libraries etc etc.		 12:05
12:05 p6bannerbot sets mode: +v Ven`
tbrowder__ when a rakudo pr is submitted, does it get the full spectest by travis? 12:24
lizmat tbrowder__: I think so 12:27
tbrowder__ thnx 12:29
anyone else having trouble getting updates from rakudo on gitgub? 12:36
or nqp? my rakudo build is failing... 12:37
lizmat last week the build failed for a few hours, but if you're up to date, all should be fine 12:38
timotimo gitgub is a fantastic typo
12:57 robertle left
AlexDaniel Geth: ver github.com/rakudo/rakudo/commit/1a...c64d2a6094 13:05
Geth AlexDaniel, version bump brought in these changes: github.com/perl6/nqp/compare/2017....gb96a0afe7
AlexDaniel Geth: ver github.com/perl6/nqp/commit/4c7f68...18b0181b79 13:06
Geth AlexDaniel, version bump brought in these changes: github.com/MoarVM/MoarVM/compare/
AlexDaniel arohragroageurohau
lizmat AlexDaniel: interesting password :-)
AlexDaniel github.com/MoarVM/MoarVM/compare/2...-gad6ab26f
I'm really confused by this: gist.github.com/Whateverable/ec73b...3ad05edf55 13:07
bisected to the moar bump mentioned above
timotimo that threw out async i/o for our std handles 13:19
jnthn releasable6: status 13:21
releasable6 jnthn, Next release in ≈2 days and ≈5 hours. 1 blocker. 1 out of 97 commits logged (⚠ 46 warnings)
jnthn, Details: gist.github.com/4a318435055ec0ad8a...f2452a6bd2
jnthn I think this'll be the first Rakudo release in 10 years that I don't have any commits in :) 13:23
Ulti do you want me to find a typo for you? >:P 13:26
jnthn My job is making those, not fixing them :P :P
AlexDaniel timotimo: aaand? 13:28
timotimo no clue :\
can you get a gdb attached and set a breakpoint to the function(s) that give that error?
AlexDaniel hmmmmm 13:29
brrt what bug are we debugging? 13:30
tbrowder__ hm, build is working again now...github must need some prune juice (OR the prune juice i just drank must have worked) 13:31
AlexDaniel brrt: not necessarily a bug, but here's what happens. There are builds of rakudo on whateverable server from 2015.12 to HEAD, and on whateverable server they work fine
timotimo "say 42" giving Unhandled exception: get_stream failed, unsupported std handle 13:32
AlexDaniel also, they (same files) were working fine on my laptop
but now 2017.05 and older no longer work
(on my laptop) 13:33
and I assume eventually they'd stop working on whateverable server also? 13:34
brrt oh, I see 13:35
AlexDaniel at least 2017.05 is buildable without any hacks for INC 13:41
13:50 brrt left
AlexDaniel Hm, I tried to golf it somewhat and it went away 13:58
though is still reproducible with a full bot…
nevermind then for now
14:19 ZzZombo_ joined 14:20 p6bannerbot sets mode: +v ZzZombo_ 14:21 ZzZombo left 14:31 |Tux| left, |Tux| joined 14:32 p6bannerbot sets mode: +v |Tux|, MasterDuke left 14:46 ZzZombo_ left, ZzZombo_ joined, card.freenode.net sets mode: +v ZzZombo_, p6bannerbot sets mode: +v ZzZombo_, ZzZombo_ is now known as ZzZombo 14:50 lizmat left 14:56 robertle joined 14:57 p6bannerbot sets mode: +v robertle 15:04 brrt joined, ExtraCrispy joined 15:05 p6bannerbot sets mode: +v brrt, p6bannerbot sets mode: +v ExtraCrispy 15:07 Zoffix joined, p6bannerbot sets mode: +v Zoffix
Zoffix tbrowder__: lizmat: no, travis doesn't run spectest. Only make test. IIRC the jobs have a 10m limit, which is barely enough just to build rakudo and make test it 15:08
15:10 Zoffix left
tbrowder__ ok, thnx, i do spectest before PR anyway 15:15
15:42 dogbert17 left 15:59 patrickb left 16:06 brrt left
tbrowder__ one more question ref PR, when merged, will it get a full spectest soon enough to revert before too much changes on master as well as nqp and moarvm? 16:13
reason for my question is my upcoming PR has a lot of code changes (mostly Pod.nqp but also Actions.nqp and Grammar.nqp 16:14
and i want to branch off the new master for another, simpler PR 16:17
on my spectests before PR i have not been doing the p5 tests nor the jvm tests because of host machine limitations 16:19
16:20 robertle left
AlexDaniel tbrowder__: I don't understand the question 16:35
tbrowder__: you should spectest your change yourself, and if it's fine then it's good to go
tbrowder__ ok, sorry to be so wordy 16:36
AlexDaniel you shouldn't worry too much about how hard would it be to revert it
16:56 Ven` left 17:00 robertle joined 17:01 p6bannerbot sets mode: +v robertle 17:18 lizmat joined 17:19 p6bannerbot sets mode: +v lizmat
lizmat so, would it be an idea to have a sub that would take a umber of awaitables and turn that into an iterator ? 18:00
if not, why ?
AlexDaniel lizmat: what about Supply.merge ? 18:04
is it similar or different from what you want? 18:05
lizmat github.com/rakudo/rakudo/blob/mast...ps.pm6#L56 :-) 18:06
I was thinking of just removing it: no tests, no spec and no documentation 18:07
(even though it has been in there since Oct 2015)
18:14 epony left
[Coke] fff 18:19
Geth rakudo: 1c3b724889 | (Elizabeth Mattijsen)++ | src/core/asyncops.pm6
Do away with the awaiterator

There was no spec, no tests, no documentation and was from before the Supplier/Supply/Awaitables refactor. Since I wrote the original version, it seems proper I'm the one to remove it as well. Shaves off about 9.5K from the .moarvm file.		 18:35
19:37 Kaiepi left
tbrowder__ i checked my spectest results and got two failures i don't think ar caused by my changes, but checking with others: t/spec/S32-list/seq.rakudo.moar and t/spec/S32-list/tail.t 20:04
lizmat those do *not* sound familiar to me 20:16
timotimo lizmat: 9.5 kilobytes? o_O 20:17
20:44 PufferBot joined 20:45 p6bannerbot sets mode: +v PufferBot
lizmat jnthn timotimo samcv any further thoughts about R#2272 ? 20:55
synopsebot R#2272 [open]: github.com/rakudo/rakudo/issues/2272 [IO] Closed handles are still readable for some time
jnthn lizmat: Didn't have time to look properly at it yet 21:06
(and certainly too tired now...)
lizmat well, I think I golfed it down a lot
it basically boils down to the decoder being in a weird state after an error in decoding 21:07
which is interpreted as having been accessed from more than one thread at the time
jnthn Yes, you're not menat to use it again after a decoding error
lizmat ok, then we need a "reset" function 21:08
jnthn No we don't
lizmat or a way of knowing it is in a weird state ?
jnthn Just make a new instance of the decoder
lizmat on a closed file handle ?
or just reset the $!decoder is what you mean?
jnthn Or that, yes 21:09
If it's closed, why are we reading from it? :)
lizmat atm you can still read from a file after it has been closed if there's stuff in the buffer 21:10
jnthn oh
lizmat R#2272
synopsebot R#2272 [open]: github.com/rakudo/rakudo/issues/2272 [IO][UNI] Closed handles are still readable for some time
jnthn that sounds like .close not invalidating stuff
lizmat yup :-)
trying $!decoder := nqp::null now 21:11
21:13 MasterDuke joined, p6bannerbot sets mode: +v MasterDuke, MasterDuke left, MasterDuke joined, herbert.freenode.net sets mode: +v MasterDuke, p6bannerbot sets mode: +v MasterDuke
jnthn iirc, there's a function that throws if we're missing the decoder, and we can probably tweak it to check the $!VMIO and complain about a read from a closed handle rather than about the thing being in binary mode 21:13
lizmat looks like "($!decoder && $!decoder := Encoding::Decoder)," is doing the trick 21:15
21:17 MasterDuke left
lizmat runna da spektezt 21:19
jnthn :) 21:25
Grr...this hotel is cute and serves nice beer but the wifi ain't the best :) 21:26
Geth rakudo: b07874e061 | (Elizabeth Mattijsen)++ | src/core/IO/Handle.pm6
Make sure the decoder is detached when closing

Fixes #2272		 21:27
synopsebot RAKUDO#2272 [open]: github.com/rakudo/rakudo/issues/2272 [IO][UNI] Closed handles are still readable for some time
Geth roast: 3253943d8c | (Elizabeth Mattijsen)++ | S32-io/io-handle.t
It already dies on the .slurp
 21:30
roast: 7e30619fba | (Elizabeth Mattijsen)++ | S32-io/io-handle.t
Add test for R#2272
synopsebot R#2272 [closed]: github.com/rakudo/rakudo/issues/2272 [IO][UNI] Closed handles are still readable for some time
tbrowder__ here's a gist with the two spectest failures i'm getting. anyone recognize them: gist.github.com/tbrowder/7fc28effe...5a823709a4 21:35
21:36 PufferBot left 21:38 PufferBot joined, p6bannerbot sets mode: +v PufferBot 21:40 Zoffix joined, p6bannerbot sets mode: +v Zoffix
Zoffix tbrowder__: yes, your rakudo isn't recent enough, I'm betting 21:40
tbrowder__ wow, it's fast moving, just updated this morning. i'll update again. thanks. 21:41
japhb .tell nine Inline::Python won't build for me at Rakudo HEAD 21:42
yoleaux japhb: I'll pass your message to nine.
Zoffix tbrowder__: are you sure you're updating from rakudo/rakudo and not from some fork of yours?
The commit on ranges went in like 1-2 days ago
tbrowder__: git remote -v
tbrowder__ no, i updated from upstream master 21:44
jnthn sleep & 21:45
evalable6 Use of uninitialized value of type Callable in numeric context
in block <unit> at /tmp/2…
jnthn, Full output: gist.github.com/c9244690d940498e7b...989c22f07b
tbrowder__ would a pull instead of a fetch made any difference?
timotimo hahaha
21:46 Kaiepi joined, p6bannerbot sets mode: +v Kaiepi
Zoffix m: dd & 21:48
camelia Callable & = Callable
tbrowder__ when i do the pull, i'm on the master branch. then i check out my new branch from that and integrate my changed code.
Zoffix ah, it's an anon state var 21:49
tbrowder__: lemme try to build master and see if that test file fails
tbrowder__ ok, thanks!
i may have made an error when integrating, but afaik i didn't touch anything like the test failures showed 21:51
Zoffix hm, nqp build fails for me t/moar/12-decodelocaltime.t fails 21:52
I mean test fails,not build. 21:53
tbrowder__ i was getting some strange msgs from github this morning and it caused problems getting a good build early on
updating moar and such
Zoffix tbrowder__: that test passes for me 21:56
This is Rakudo version 2018.08-99-gb07874e06 built on MoarVM version 2018.08-71-gccf3dd373
Full log: gist.github.com/zoffixznet/d1b6532...5d4e8110e9
tbrowder__ This is Rakudo version 2018.06-586-g0a416da built on MoarVM version 2018.08-63-ga8dfd47 21:57
hm, the garden needs tending...back later
21:59 Zoffix left
timotimo tender lawn and carrots 21:59
lizmat tbrowder__: my rakudo is at 2018.08-99-gb07874e 22:06
tbrowder__ i just updated again and picked up some more changes, rebuilding master now (non of my code) to see what i get with ./perl -v 22:09
Rakudo version 2018.06-588-gb07874e built on MoarVM version 2018.08-63-ga8dfd47 22:13
weird
at least moar is right
lizmat has made some progress on R#2272. but is too tired now to work out the final bits 22:14
synopsebot R#2272 [open]: github.com/rakudo/rakudo/issues/2272 [IO][UNI][testneeded] Closed handles are still readable for some time
lizmat sleep&
tbrowder__ ok, got masters, please tell me what magic git incantation do i need to get the "real" master HEAD? 22:38
ok, git masters, please tell me what magic git incantation do i need to get the "real" master HEAD?
my configure line is "Configure.pl --backends=moar --gen-moar --gen-nqp" which installs into ./install 22:42
timotimo tbrowder__: you can --gen-moar=master --gen-nqp=master 22:52
22:56 MasterDuke joined, p6bannerbot sets mode: +v MasterDuke, MasterDuke left, MasterDuke joined, herbert.freenode.net sets mode: +v MasterDuke, p6bannerbot sets mode: +v MasterDuke
MasterDuke timotimo: around? 22:58
timotimo a squared 22:59
tbrowder__ timotimo: thanks, sounds like a good plan!
MasterDuke timotimo: heh. did you happen to catch the link i sent a day or two ago with info from the debug server? 23:01
timotimo yeah, it looked so innocuous :(
i tried to do the "non-stop gdb + debugserver" thing myself the other day and couldn't get it to work :( 23:03
MasterDuke huh, i just did what you suggested and it worked
timotimo i did what i suggested and it hung :( 23:05
MasterDuke now, i haven't actually tried to resume anything... 23:06
timotimo me neither
MasterDuke is there a MVM_gc_mark_thread_unblocked? 23:07
timotimo yes 23:11
MasterDuke do i need to resume in the debugserver first?
timotimo wellllll
resuming in the debugserver will try to signal a thread that's inside the mark_thread_blocked piece thingie 23:12
all you need to do™ is reset the gc_status to what it was before, which is most probably 0
MasterDuke i never actually did anything with gc_status 23:13
timotimo oh?!
i would have expected the debugserver would have refused doing anything for you if you left it at 0 23:14
so maybe that's the mistake i made
MasterDuke so you can't think of any other info to get before i try resuming? 23:17
timotimo wellllllllllll 23:18
MasterDuke heh
timotimo what you can do, you see, is plug objects manually into the debug server's handle table
meaning even if you don't have a lexical to hold the thing you're interested in 23:19
MasterDuke but it was a lexical, right?
timotimo i'm not sure any mor
MasterDuke gist.github.com/MasterDuke17/91c58...5994973fc6 23:20
i found it in the output of `all lex 1` 23:21
89 QAST::Op $op concrete
timotimo OK, what errored again? 23:30
MasterDuke $op.push 23:31
`$op.push(QAST::Stmts.new(` that's the line it said 23:32
`Cannot look up attributes in a VMNull type object` 23:33
timotimo hum, but the push method isn't on the call stack? 23:34
can you show the whole code?
oh
line numbers in the debugserver lie!
MasterDuke no, the line number is from before using the debugserver 23:35
23:36 epony joined, p6bannerbot sets mode: +v epony
MasterDuke i'm broken at the nqp op i inserted just before that line 23:38
can i try stepping in the debugserver, or do i need to call *unblock in gdb?
23:39 ZzZombo_ joined 23:40 p6bannerbot sets mode: +v ZzZombo_ 23:41 ZzZombo left
MasterDuke timotimo: this is the diff of my current changes gist.github.com/MasterDuke17/1c45e...d49eabd36c 23:43
timotimo ah, asec_n(2e1) 23:44
btw, the whole entire thing has only one single "line" 23:45
er, actually
maybe not
do you have the "self"?
MasterDuke in the debug output? 23:46
timotimo access to it somehow, yeah
MasterDuke 55 Perl6::Optimizer self concrete
same frame as $op
timotimo oh
why didn't i think of this
you can totally step through the moarvm interpreter in gdb
it's not as nice an experience at all 23:47
but it'll step op by op
can i get more output from your debug session? including the "all lex" one?
MasterDuke gist.github.com/MasterDuke17/f2c0a...a1f46cdafe 23:48
timotimo line 2342 is just an inner { } thing? 23:49
MasterDuke ? 23:50
timotimo well, what constitutes frame 0?
MasterDuke the nqp::asec_n ? 23:51
timotimo i'm not sure if Frame 0 is just an inner block inside Frame 1
but it most probably is 23:52
just from the else { ... }
can you tell me which commit you're based on so i can apply the patch myself? 23:55
MasterDuke de4e82259be36ac9c97b56d594591ea796a1b131
timotimo patch unexpectedly ends in middle of line 23:56
MasterDuke i just put the .patch file in the gist instead of copy/pasting 23:57
timotimo found a non-native Int 23:58
Cannot invoke this object (REPR: P6opaque; NQPMu)
at gen/moar/Perl6-Optimizer.nqp:2342 (blib/Perl6/Optimizer.moarvm:)
this is the right crash?
MasterDuke yep

Powered by Cro and the Raku® Programming Language.

Please report any issues / comments / feature requests as an issue on App::Raku::Log.

Thank you!