jnthn damm, this chilli ölkorv is hot 00:03
Must be designed for drinking crappy lager where you don't care about the taste or something... 00:04
469 :) 00:07
00:30 ssutch joined 00:49 ssutch joined 00:53 ssutch joined 01:02 BenGoldberg joined
jnthn Up to line 664 by now. 01:29
Think I'm too tired to do more today. Goal for tomorrow: get past 1000. :) 01:34
diakopter ..so many diakopter thinkos... :( 02:34
moritz gist.github.com/moritz/f890a56c13ddc849e537 # moar assertion failure while building rakudo/moar-support 02:45
yesterday I also had one while building NQP
in both cases, simply typing 'make' again "fixed" the problem
diakopter hm
moritz erm no, in rakudo re-making doesn't fix it
fwiw dynext/libperl6_ops_moar.so exists 02:46
dynext/libperl6_ops_moar.so: ELF 64-bit LSB shared object, x86-64, version 1 (SYSV), dynamically linked, BuildID[sha1]=0x74e8f1e273673ba68fff7037a0f28df1eac08e21, not stripped
tadzik well, it doesn't build for me at all 03:05
diakopter tell it to make harder 03:08
tadzik gist.github.com/tadzik/7508563 03:10
ok, cleaning linenoise dir helped 03:16
moritz make: *** No rule to make target `harder'. Stop. 03:25
03:52 ssutch joined
FROGGS +++++++++++++++++++++++++++++++++ 07:19
++++
++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++ 07:21
###########
´´´´´´´´´´´´´´´´´´´´
´´´´´++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++ 07:22
TimToady any solar flares lately? 07:27
or maybe we can blame the Leonids... 07:30
It would be exremely funny if the first evidence of ETs showed up as noise on an IRC channel... 07:31
FROGGS I'm sorry 08:10
seems like my youngest is interested in computers :/
moritz: when was the last time you could "build" rakudo@moar? 08:13
looks like I can build and hit: No registered operation handler for 'p6captureouters' 08:30
moritz: wrt "failed to load library 'dynext/libperl6_ops_moar.so'", I would think a `make realclean` and rebuild in MoarVM would help here... but I thought we did that a few weeks ago? 08:32
I am working on that now: 09:02
perl Configure.pl --backends=jvm,moar --prefix=../nqp/install && make m-install
===SORRY!===
nqp-j is too old
wait, that one is correct 09:10
moritz linking moar 10:27
./libmoar.so: undefined reference to `MVM_repr_bind_key_boxed'
collect2: ld returned 1 exit status
when rebuilding moar
FROGGS did you reconfigure?
because that happens when then src/gen/config.c is old 10:28
and maybe you are even missing about 10 commits? jnthn++ fixed it yesterday
moritz do I need to rebuild nqp after rebuliding moar? 10:29
or can I shoot straight for rakudo?
FROGGS you can usually skip nqp
moritz ok, after realcean, git clean -xdf and reconfiguring, moar build worked again 10:30
FROGGS phew
btw, is the following supposed to gen-moar? perl Configure.pl --backends=moar --gen-nqp --gen-moar --make-install
(in rakudo)
ahh, there is no gen-moar yet, right? 10:31
no, there is...
moritz only in nqp
FROGGS yeah... 10:33
timotimo FROGGS: when the youngest get interested in computers, it's time to get a filesystem that can revert any change made to it in the last 12 hours, granually :P 10:43
FROGGS timotimo: that is a good idea *g*
timotimo what do i have to do for rakudo-moar to build the dynext thingie? 10:54
FROGGS pull everything, make realclean in moarvm, reconfigure and build 10:55
timotimo mhm
FROGGS moritz: btw, I see that gen_parrot is handled in rakudo also... does that make sense? why don't we just pass a --gen-parrot to nqp? 10:56
moritz: I did the same for gen-moar now
umm, nqp HEAD does not build on moar :/ 11:12
timotimo locally it works 11:13
FROGGS it fails building QRegex on my box 11:14
timotimo make -j4 all install 659,28s user 10,75s system 264% cpu 4:13,68 total 11:15
interestingly, it doesn't seem to want to accept my nqp-j version
FROGGS timotimo: I bumped nqp revision right now, so you have to reconfigure/rebuild nqp 11:18
timotimo i did, though
gcc: error: dynext/perl6_ops.o: No such file or directory
i'm not sure what i'm doing wrong, but maybe i should just wait until more work is done on the whole thing
FROGGS timotimo: can you paste the build output of rakudo? 11:20
timotimo has reconfigured nqp and is rebuilding 11:24
FROGGS: gist.github.com/timo/3605ab3ba130d90c563e 11:34
maybe the warning in the configure output has something to do with it ... 11:35
FROGGS timotimo: what is: nqp-m --version | wc 11:39
err
timotimo: what is: nqp-m --show-config | wc
timotimo 102 319 8192
[2] 11089 segmentation fault (core dumped) nqp-m --show-config |
:)
that's not very good
FROGGS timotimo: that commit fixed it for me: github.com/MoarVM/MoarVM/commit/54...5c7996a009 11:40
timotimo: do you have time to generate a backtrace?
timotimo sure. 11:41
FROGGS k
cat install/bin/nqp-m
timotimo same url
exec /home/timo/build/rakudo/install/bin/moar --libpath=/home/timo/build/rakudo/install/languages/nqp/lib /home/timo/build/rakudo/install/languages/nqp/lib/nqp.moarvm "$@" 11:42
oh, that could be wrong
i think my moar prefix is inside nqp/
FROGGS no, perfet
(i think)
ahh
timotimo gimme a minute.
FROGGS then: /home/timo/build/rakudo/install/bin/moar --version
phew 11:48
my --gen-moar in rakudo seems to work... the segfault I got is something I "patched" locally, but this is not committed
forgot about that patch/hack though 11:49
timotimo i know that feeling :P
yesterday when i was fiddling around with regex codegen and such i ran into lots of unstable problems
FROGGS yeah, that sounds that regex fiddling :o) 11:50
moritz: what do you think about gist.github.com/FROGGS/fdbfe209806fba0587ad ?
moritz FROGGS: should also include a $backends{moar} = 1 if defined $gen_moar; 11:51
or something along those lines, at least 11:53
FROGGS ahh, k 11:54
11:55 tgt joined
moritz note that the parrot case is more complicated, because we pull copmiler flags etc. from the parrot configuration that aren't available from the NQP configuration 11:55
FROGGS ahh, I see 12:20
12:20 diakopter joined 12:52 _ilbot joined
FROGGS jnthn: I still need this patch to avoid an explosion in the gc: gist.github.com/FROGGS/603692fde738639fc0d3 13:04
moritz FROGGS: and reason you didn't include the $backends{moar} = 1 if defined $gen_moar; line? 13:05
FROGGS: also I'm not convinced the $impls{moar}{ok} = 1; line is right 13:06
FROGGS jnthn: without that patch I even got once or twice the message about something invalid and a mention of fromspace
moritz FROGGS: it should conclude that when it successfully reads the nqp-m config, and the version is new enough
FROGGS ahh, hmmm
ahh, now I know what you mean 13:07
jnthn FROGGS: That's a slightly weird patch to need... 13:12
FROGGS jnthn: that is why it is not committed...
and the funny thing is: why do only I need that patch? 13:13
maybe because of gcc (Ubuntu/Linaro 4.8.1-10ubuntu8) 4.8.1?
jnthn FROGGS: Does gist.github.com/jnthn/7513289 help at all? 13:18
FROGGS /home/froggs/dev/nqp/install/bin/moar --libpath=src/vm/moar/stage0 src/vm/moar/stage0/nqp.moarvm --bootstrap --setting=NULL --no-regex-lib --target=mbc \ 13:20
--output=gen/moar/stage1/nqpmo.moarvm gen/moar/stage1/nqpmo.nqp
Segmentation fault (core dumped)
:o(
jnthn bt? 13:21
oh, hang on, that SEGVs it for me too...wtf
FROGGS gist.github.com/FROGGS/dcf494aecec7506fdd79
you give me untested code? wtf :P 13:22
jnthn It was too simple not to wok :P
uh, work
FROGGS yeah, one of those :o) 13:23
FROGGS .oO( To work or not to work? )
jnthn Oops. It looks like somewhere is pushin' temproots and not poppin' em late 13:27
13:27 moritz joined
jnthn Having 74 of them when we should have 1 is a sure sign of fail 13:28
FROGGS jnthn: MoarVM/src/core/threads.c:104: MVM_gc_root_temp_push(child_tc, (MVMCollectable **)&ts->thread_obj); 13:32
I don't see a pop there
jnthn Tht looks odd, but there's no threads involved here... 13:35
FROGGS isn't it more likely that the temproots would be popped later? 13:37
I don't see another push without the pop
jnthn FROGGS: An MVMROOT with a return in it would also cause issues 13:39
FROGGS ahh, true 13:40
jnthn oh, which is what my patch did
Which is why it busted stuff
FROGGS hehe
jnthn Darn, and when I take away my patch, we never hit the problem of too many temp roots. 13:47
FROGGS here are returns in MVMROOT... MoarVM/src/io/fileops.c:192: MVMROOT(tc, dot, { 13:49
(I did that btw)
jnthn ah...that should be fixed...
FROGGS doing that atm 13:50
jnthn Back in loadbytecode.c, does valgrind complain at all?
FROGGS with unmodified moarvm? 13:51
jnthn yeah 13:52
FROGGS hmmm 13:53
jnthn FROGGS: Could also try gist.github.com/jnthn/7513691
FROGGS I fixed the one in fileops.c, and it compiles further... 13:54
that was the only return in MVMROOT btw 13:56
nqp compiled successfully 13:57
compiling rakudo now
FROGGS scratches his head 14:01
jnthn: it compiled till the NYI
dalek arVM: f2bbadf | (Tobias Leich)++ | src/io/fileops.c:
never ever return from inside a MVMROOT
 14:02
jnthn FROGGS: yay
OK, guess I should look at the NYI...
FROGGS sorry for the distraction :o) 14:03
jnthn np, nice it's fixed
FROGGS k, there is no other return between push and pop 14:06
jnthn \o/ 14:13
FROGGS ? 14:14
jnthn fileops.c 14:15
src\io\fileops.c(211) : error C2065: 'last' : undeclared identifier
:(
FROGGS uhh
there is no `last` in C, ehh?
jnthn break :) 14:16
FROGGS ahh 14:17
dalek arVM: 1b6024c | (Tobias Leich)++ | src/io/fileops.c:
s/last/break/
 14:19
FROGGS jnthn: btw, that does not make sense at all 14:22
because the return was in _WIN32 code 14:23
jnthn oh...
FROGGS I'll paste a bt once I hit it again 14:24
bbl
dalek arVM: e15a28a | jnthn++ | src/core/frame.h:
Mark a couple more things MVM_PUBLIC.

Needed for Rakudo's ops.
14:32 woolfy1 joined
dalek arVM: 62964cc | jnthn++ | src/core/interp.c:
decont should cope with nulls.
 14:40
14:42 woolfy joined, lizmat_ joined 14:43 FROGGS[mobile] joined
jnthn Up to line 2794 14:47
FROGGS wow!! 14:50
jnthn++
jnthn At Frame 2, Instruction 37, op 'istype', operand 1, MAST::Local of wrong type (7) specified; expected 8 15:04
urgh
oh, maybe it's just a small thing... 15:07
15:21 Woodi joined
FROGGS[mobile] a short perhaps? 15:25
jnthn :P
Fixed that one now.
15:32 BenGoldberg joined 15:36 Woodi joined
FROGGS Incompatible MROs in P6opaque rebless 16:18
is that so?
jnthn That means you're some commits behidn somewhere... :) 16:19
Probably NQP ones
FROGGS O.o
ahh, moarvm 16:20
ahh, yes 16:46
dalek arVM: 3c262e6 | jnthn++ | src/ (3 files):
Fill out some HLL type-mapping logic.

So things crossing MOP-in-NQP to Perl 6 boundary get the right types.		 16:54
arVM: f92265c | jnthn++ | src/core/coerce.c:
Implement bigint boolification.
jnthn Getting up to around line 3733 now 16:59
The Numeric role is the next place we fail
Error while compiling op chain (source text: "=="): No registered operation handler for 'chain' 17:00
nwc10 build of HEAD now fails like this: 17:04
moar: 3rdparty/libuv/src/unix/loop.c:150: uv__loop_delete: Assertion `!((((*(&(loop)->active_reqs))[0]) == (&(loop)->active_reqs)) == 0)' failed.
make: *** [blib/Perl6/BOOTSTRAP.moarvm] Aborted
timotimo Incompatible MROs in P6opaque rebless <- /me, too 17:09
jnthn timotimo: That was fixed in nqp commit ce7482c 17:12
timotimo ah, okay
nwc10 my assistant is preventing me from seeing the screen to update and rebuild 17:14
she wants me to "play barbapapa" 17:15
jnthn
.oO( Operation Barbapapa )
 17:16
nwc10 "play" being the real word that needs the scare quotes
now: Error while compiling op chain (source text: "=="): No registered operation handler for 'chain' 17:27
massive backtraace
and an assertion at the end: 17:28
moar: 3rdparty/libuv/src/unix/loop.c:150: uv__loop_delete: Assertion `!((((*(&(l
oop)->active_reqs))[0]) == (&(loop)->active_reqs)) == 0)' failed.
make: *** [CORE.setting.moarvm] Aborted
jnthn nwc10: Yup, the chain thing is what I'm fixing at the moment
nwc10 OK, so that's current and expected 17:29
jnthn At Frame 3, Instruction 37, op 'prepargs', operand 0, MAST::Local of wrong type (8) specified; expected 7 17:34
wtf, I don't even...
FROGGS nwc10: the libuv stuff always shows up when we print to stderr 17:40
nwc10 OK
jnthn oh, it helps to use the correct set of flags... 17:43
4465 17:49
FROGGS that is awesome :o) 18:03
jnthn 4495 now. We survive our first enum, but not our first subset. 18:05
FROGGS damn it, I wanna know how long stage parse will take :o) 18:07
jnthn ages... :S 18:09
Though maybe it's 'cus of all the warnings being spewed... 18:11
FROGGS yeah, hopefully :/ 18:14
jnthn Well, don't forget we did none of the interesting optimization stuff yet in Moar. 18:15
FROGGS okay, it dies after 39s on my box 18:16
jnthn Aww, fixing the missing p6var only gets me to 4500.
nwc10 awll
can always take the Klortho-disapproved approach to optimisation:
#11912 Evidently it's important to you to get the wrong answer as quickly as possible.
jnthn I'll bet profiling will reveal a bunch of things too :) 18:17
nwc10 ln -s /usr/bin/true /usr/bin/perl6
jnthn I mostly care about getting it to *work*.
(for the moment) 18:18
dalek arVM: 5bf1995 | jnthn++ | src/strings/ops.c:
Allow join on array of native strings.
 18:30
FROGGS was that 5038? 18:31
nwc10 bah, I haven't finished building the pervious version and the goalposts moved again :-) 18:32
jnthn Latest thing we choke on is p6return
FROGGS nwc10: same here :o)
ahh damn, an optimized moar segfaults... I wanted to compare the timings so far 18:36
takes 1m11s atm to get to p6return
jnthn I'm not sure where exactly that p6return is, fwiw 18:37
A when in a role would do it 18:38
FROGGS I've have read something about role_body 18:39
jnthn Oh, it's probably the one in Blob.unpack
Meaning we've parsed up to around 5163
FROGGS ahh 18:41
'`when` in a role'
now I parse your message correctly :o)
jnthn uh, yes :)
FROGGS that would fit, because of when_handler_helper or the phaser stuff 18:42
jnthn Anyway, seems we're beyond line 5000
FROGGS jnthn++ # :o)
timotimo the next 5000 will also take about a day, then? :)
FROGGS less
there won't be spectacular things in the range of 5000 .. 10_000 I guess 18:43
jnthn Well, don't jinx it
FROGGS yeah, true 18:44
jnthn There's always the chance we hit a nasty GC bug, for example.
FROGGS damnit
now he said it -.-
:P
FROGGS .oO( Beatlejuice Beatlejuice Beatlejuice! )
jnthn OK, so where on earth is the next error coming from... 19:07
FROGGS cannot stringify this 19:08
Error while constructing error object:Could not locate compile-time value for symbol Undeclared
this?
$*W.throw($var, ['X', 'Undeclared'], symbol => $varast.name(), suggestions => @suggestions);
rakudo/src/Perl6/Grammar.nqp:284 19:09
jnthn Yeah 19:10
But what var is it on about
jnthn is trying to work out where we get to
FROGGS I'm trying to say its name
jnthn That's a clever way
I'm shoving BEGIN nqp::say('here 1') s into CORE.setting... 19:11
FROGGS ahh, nice trick
jnthn Whoa, we make it *past* the one I shoved in at line 7000 or so
FROGGS \o/
so, done today? *g* 19:12
jnthn oh...
It's a legit error too I think
method new($block, Mu :$infinite) {
my Mu $new := nqp::create(self);
nqp::bindattr($new, GatherIter, '$!coro', $coro);
FROGGS Cannot invoke null object
jnthn I see no $coro 19:13
FROGGS varast is null?
moritz ... and the other backends don't complain?
FROGGS the bang is missing
jnthn moritz: It uses conditonal comp 19:14
moritz oh.
jnthn moritz: And theere wasn't a block for moar
I just shoved in a my $coro = die 'NYI' or so
for moar
FROGGS Backtrace.pm:40 is such a candidate too 19:16
src/core/Exception.pm:159 19:17
shall I fix these?
jnthn candidate for...? 19:18
FROGGS a missing moar variant
it generates invalid code
moritz nqp-m: nqp::backtrace(NQPMu)
camelia nqp-moarvm: OUTPUT«Error while compiling op backtrace (source text: "nqp::backtrace(NQPMu)"): No registered operation handler for 'backtrace'␤frame_name_1109␤»
FROGGS there are branches for #? parrot and #? jvm
moritz FROGGS: it won't even copmile
jnthn ah, ok
yeah, go for them
Our next issue is missing nqp::p6argvmarray 19:19
Used in a role
iiuc, that means we're at 8247
FROGGS nqp-m: nqp::readfh() 19:21
camelia nqp-moarvm: OUTPUT«Error while compiling op readfh (source text: "nqp::readfh()"): No registered operation handler for 'readfh'␤frame_name_1109␤»
FROGGS :/
jnthn oops, it's 8:20pm already and I didn't dinner
FROGGS uhh, quick!
nqp-m: nqp::writefh() 19:22
camelia nqp-moarvm: OUTPUT«Error while compiling op writefh (source text: "nqp::writefh()"): No registered operation handler for 'writefh'␤frame_name_1109␤»
jnthn yay, now FROGGS++ has found things he can hack on the advance state of Rakudo on Moar ;)
FROGGS :o)
nqp-m: nqp::tellfh() 19:23
camelia nqp-moarvm: OUTPUT«Error while compiling op tellfh (source text: "nqp::tellfh()"): Arg count 0 doesn't equal required operand count 2 for op 'tell_fh'␤frame_name_1109␤»
FROGGS good
jnthn Tuit supply has been good this weekend, but tomorrow I need to worry about worrying $dayjob thing...
FROGGS yeah, mondays... 19:25
jnthn Well, I was meant to have 2-3 days for Perl 6 this week...alas, I need to help avert various disasters... 19:26
Happy we're chewing through parsing lots of CORE.setting with Moar by now, though... 19:29
FROGGS yeah, I have not expected to get that far 19:30
well, you know what I mean :o) 19:31
[Coke] jnthn++
FROGGS how does printfh differ from writefh? 19:32
jnthn writefh is the one that takes a buf-y thing, iirc 19:33
FROGGS ahh, yes, now I see it too
jnthn shop; bbiab 19:36
dalek arVM: 35016eb | (Tobias Leich)++ | / (8 files):
implement read_fhb which return an int8-array
 20:35
arVM: 6d49b5d | (Tobias Leich)++ | / (8 files):
implement write_fhb
 20:48
jnthn FROGGS: There's a little cluster of unimpl ops around 8250, turns out...
But given what we hit right after what I just fixed, I'm pretty confident that's how far we are. 20:49
FROGGS I guess you mean p6bindattrinvres and p6scalarfromdesc 20:50
jnthn and p6list 20:51
FROGGS hmm, no, both appear before
jnthn but yeah, these.
FROGGS ahh, k
jnthn But we get away with it unless it's in a role...
We'll hit a bunch of others at stage mast I guess :) 20:52
FROGGS :o)
21:00 ssutch joined
FROGGS k, so p6scalarfromdesc is above my head 21:30
timotimo is p6list getting tackled yet? 21:32
i wonder what that op has to do
dalek arVM: 0806222 | jnthn++ | src/gc/wb.h:
Make write barrier hit function public.

Otherwise, we can never MVM_ASSIGN_REF in external code.		 21:33
timotimo there's probably much more to p6list that's not visible in the implementation i'm looking at right now (on jvm) 21:35
21:35 _sri joined
jnthn Just pushed p6list 21:37
timotimo oh :)
jnthn Working on 'p6bindattrinvres
Which was easy. Now for p6scalarfromdesc. 21:42
nwc10 On Linux HEAD bombs out at 21:56
Error while compiling op p6scalarfromdesc: No registered operation handler for '
p6scalarfromdesc'
which I guess is what you'd expect
jnthn aye 21:57
FROGGS nwc10: which is at line 8276 of 16956 :o) 21:58
nwc10 nqp-m: say(8276/16956) 22:00
camelia nqp-moarvm: OUTPUT«0.488087␤»
FROGGS aww
jnthn D'oh, p6bindattrinvres had a couple of thinkos, masked by the missing p6scalarfromdesc since that was used in its args... 22:01
FROGGS :/ 22:05
where are we now? do you know that? 22:19
jnthn Not really 22:20
FROGGS k
jnthn Something tries to actually use p6store... 22:22
FROGGS how bad of it! :P
jnthn Trying to work out where we are as I think it may be "more than half way" 22:24
FROGGS well, the only p6store that appears literally is at line 15861 22:25
but that can't be it :o) 22:26
jnthn no, but = is implemented in terms of it :)
FROGGS yeah, put there are some of them, no?
FROGGS tweaks gen-cat 22:27
jnthn ah, I think I know what's causing it 22:28
Well, what's triggering p6store I mean
yeah, it's 22:29
constant i = Complex.new(0e0, 1e0);
Which is on line 10,374
FROGGS yay!
seen it :o) 22:30
then my gen-cat should print 'src/core/Complex.pm'
jnthn nqp-m: say(10374/16956) 22:31
camelia nqp-moarvm: OUTPUT«0.611819␤»
jnthn Guess we got my target of "past line 1000" :) 22:32
FROGGS hehe
I think so too :P
jnthn Anyways, p6store probably needs real work. :)
FROGGS 10k lines within a day... that is pretty awesome :o) 22:35
jnthn Oh...I wonder if we should just emit p6store as a number of instructions rather than trying to ext-op it... 22:36
yeah, that'd make it a load more manageable
And when we get a specializer it can probably eliminate an entire branch of it...
FROGGS my modified gen-cat works fine... 22:40
which prints the name of the cat-ed file we are in, except for the first
yes, it prints Complex.pm :o) 22:46
jnthn I'll probably look at p6store tomorrow evening. 22:48
Unless I decide to just do a crazy number of hours on the $dayjob thing tomorrow to try and give it a firm shove off my plate... :) 22:49
FROGGS I guess I'll look more into S11 tomorrow again 22:50
jnthn Cool
Good to keep lots of things moving along.
FROGGS like trying to load two different versions of a module
23:54 benabik joined

Powered by Cro and the Raku® Programming Language.

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

Thank you!