Camelia, the Perl 6 bug

IRC log for #moarvm, 2013-09-07

| Channels | #moarvm index | Today | | Search | Google Search | Plain-Text | summary

All times shown according to UTC.

Time Nick Message
00:13 FROGGS joined #moarvm
00:16 benabik joined #moarvm
00:38 FROGGS joined #moarvm
00:58 benabik joined #moarvm
01:08 dalek MoarVM/gcorch: d77cd3a | diakopter++ | / (11 files):
01:08 dalek MoarVM/gcorch: wip
01:08 dalek MoarVM/gcorch: review: https://github.com/MoarVM/MoarVM/commit/d77cd3add9
01:09 diakopter benabik: ^ nice commit message ;)
01:09 benabik heh
01:10 diakopter you like the diff? lots of changes.
01:11 diakopter and lots of breakage currently
01:11 benabik gcorch?
01:12 diakopter gc orchestration
01:12 diakopter coordination of threads
01:13 diakopter jnthn will be *much* happier with the new stuff.
01:13 diakopter I kindof bungled it previously
01:13 diakopter now it's starting to look like real software
01:13 diakopter orchestration.c that is.
01:13 diakopter run.c still needs work
01:13 diakopter orchestrate.c I mean
01:14 diakopter bbl
01:35 benabik joined #moarvm
03:29 ingy benabik: ping
03:29 benabik ingy: pong
03:30 ingy :D
03:30 ingy https://gist.github.com/ingydotnet/6472496
03:30 ingy look at that a sec
03:30 ingy would you agree that api.github.com is broken
03:30 benabik ETag not changing?
03:31 benabik That's less than ideal.
03:31 ingy or am I broken?
03:34 benabik Yeah, no that's broken.
03:35 benabik If the Etag hasn't changed, the content shouldn't have changed.
03:41 FROGGS joined #moarvm
03:41 ingy right!
03:42 ingy the content is completely correct, and different, but with the same etag
03:49 JimmyZ Good morning
03:58 diakopter ingy: I bet if you report it they'll fix it quickly
04:01 ingy diakopter: I'm not so sure
04:01 ingy Thanks for getting in touch. Since our ETags use strong validation [1], it seems like they should indeed change as you've described. Thanks for pointing this out.
04:01 ingy I'm adding this to our Wish List of enhancements. That list is getting quite long at the moment, so I don't know for sure when/if our small API team will be able to ship this enhancement. But, it's definitely on our radar.
04:01 ingy Thanks again for the suggestion!
04:01 ingy this was about another less important but related bug
04:11 ingy to be honest, nothing about programming seems to irk me as badly and often as the github bug reporting process
04:12 FROGGS joined #moarvm
04:12 ingy I've screamed at them (via email and irc and even their reporting system) many times about how infuriating it is to work through their closed system
04:13 ingy look at this: https://github.com/contact
04:13 ingy would you even think that is a bug reporting system?
04:14 ingy I love the new "I need help with an order from shop.github.com" checkbox in the middle
04:15 ingy makes me want to go to SF with a sawed-off shotgun :\
04:15 * ingy needs a cocktail :[
04:15 ingy brb
04:17 * ingy wonders if a Rolling Rock tallboy counts as a cocktail…
04:17 ingy *cheers*!
04:31 JimmyZ ../moarvm nqp.moarvm --target=mbc t/nqp/01-literals.t
04:31 JimmyZ This representation (SCRef) does not support elems
04:57 * JimmyZ doesn't know why make selftest all fails only on x86
05:06 benabik Memory layout change hitting a GC bug?
05:17 FROGGS joined #moarvm
05:53 not_gerd joined #moarvm
05:53 not_gerd o/
06:21 JimmyZ \o
06:26 FROGGS o\
06:42 dalek MoarVM: 175faaa | (Gerhard R)++ | build/Makefile.in:
06:42 dalek MoarVM: Implement MVM_platform_cpu_count()
06:42 dalek MoarVM:
06:42 dalek MoarVM: Untested on non-windows (as in, might not even compile).
06:42 dalek MoarVM:
06:42 dalek MoarVM: If so, please provide a fix or remove the offending file
06:42 dalek MoarVM: from PLATFORM_POSIX in build/Makefile.in
06:42 dalek MoarVM: review: https://github.com/MoarVM/MoarVM/commit/175faaae0a
06:42 not_gerd oO( I like the smell of breaking master in the morning )
06:45 not_gerd ups..
06:45 not_gerd forgot to add the new files
06:46 dalek MoarVM: 78b5015 | (Gerhard R)++ | src/platform/ (4 files):
06:46 dalek MoarVM: Actually add new files
06:46 dalek MoarVM: review: https://github.com/MoarVM/MoarVM/commit/78b5015182
06:54 JimmyZ ha.
07:36 * FROGGS builds moarvm on a fresh windows 7 box
07:58 JimmyZ x86 box?
08:04 FROGGS inter::progs -        Determine what C compiler and linker to use...Compilation
08:04 FROGGS failed with 'cl'
08:04 FROGGS Command failed (status 256): C:\Perl64\bin\perl.exe Configure.pl --optimize --pr
08:04 FROGGS efix="C:/MoarVM/nqp-cc/nqp/install"
08:04 FROGGS no x64
08:04 FROGGS can't configure parrot
08:05 FROGGS I'll try my win xp x86 box later, but this one has other problems I need to fix first
08:06 FROGGS maybe my windows 7 box is missing some windows sdk?
08:32 not_gerd_ joined #moarvm
08:42 dalek Heuristic branch merge: pushed 17 commits to MoarVM/validation by gerdr
09:10 not_gerd Bytecode validation error: callsite expects 65532 more positionals
09:10 not_gerd so, I guess I should look into named args ;)
09:27 not_gerd why isn't MVM_CALLSITE_ARG_FLAT_NAMED == MVM_CALLSITE_ARG_FLAT | MVM_CALLSITE_ARG_NAMED
09:28 not_gerd jnthn: ^^
09:33 jnthn 'cus you process a flattening named arg more like a positional (one slot in the *args) rather than a named (two lots in the *args, a name constant and a value).
09:33 jnthn *slots
09:34 not_gerd that's highly relevant to what I'm doing
09:35 jnthn And changing it would mean you can't look for MVM_CALLSITE_ARG_FLAT and know there's a name followed by a value in the args buffer.
09:35 JimmyZ btw: make selftest all failes  only on x86. make test and make nqptest not
09:35 jnthn JimmyZ: Then, some debugging or bisecting is needed...
09:36 JimmyZ e
09:39 not_gerd jnthn: I'm assuming we don't generate NAMED | FLAT (more)
09:39 not_gerd so we have regular and FLAT_NAMED args taking 2 slots and NAMED args taking 1 slot
09:39 jnthn um, wrong way around, no?
09:40 not_gerd yes ;)
09:40 jnthn FLAT_NAMED = a positional that needs flattening later, NAMED = two slots...
09:40 * not_gerd should probably just peek at args.c
09:40 jnthn And no, you won't get FLAT_NAMED and NAMED set on the same thing.
09:41 jnthn Well, shouldn't :)
09:43 not_gerd I'll be able to tell you if we do soon-ish
09:43 jnthn :)
09:47 not_gerd just to make extra sure: FLAT_NAMED does contribute to num_pos?
09:47 jnthn not_gerd: Correct
09:48 jnthn Reading bytecode.c around 492 may help clear things up
09:48 not_gerd I'm probably good
10:16 donaldh joined #moarvm
10:25 FROGGS jnthn: except vs2012express, git and activeperl, do I need something else to build parrot on win7 ?
10:25 jnthn FROGGS: Think that's what I use...though maybe I'm on the 2010 toolchain...
10:25 FROGGS yeah, maybe it is just that...
10:26 jnthn Well, I think it's 2012 on my laptop and it works...
10:26 FROGGS hmmm
10:26 jnthn x64 probably though
10:26 FROGGS correct
10:27 FROGGS using an 64bit activeperl
10:27 FROGGS (running configure in the vs developer console)
10:28 jnthn aye, that's what I use...
10:28 jnthn Are you getting issues?
10:28 FROGGS yeah, see clogs from about 10am
10:28 jnthn huh, I have to read...
10:28 Guest1337 joined #moarvm
10:29 jnthn oh, huh...
10:29 jnthn FROGGS: I've sometimes seen that happen randomly, and a second attempt works...
10:29 jnthn FROGGS: But, type
10:29 jnthn cl
10:29 jnthn And hit enter
10:29 jnthn Does it find something?
10:29 FROGGS I get its help msg, yeah
10:29 jnthn ok, so it's not that it's missing...
10:30 FROGGS I think I'm debugging parrot's build system after lunch, so I know what commands it is trying to run
10:31 jnthn I think it's meant to spit out a file saying how it failed
10:31 jnthn Try Configure with --verbose maybe
10:32 FROGGS k, thanks
10:36 dalek MoarVM/validation: b87a6e3 | (Gerhard R)++ | src/core/validation.c:
10:36 dalek MoarVM/validation: Validate number of positional arguments
10:36 dalek MoarVM/validation: review: https://github.com/MoarVM/MoarVM/commit/b87a6e3143
10:38 FROGGS jnthn: what is th equivalence of LC_ALL=C for windows?
10:39 FROGGS I think it tries to link to 32bit libs for x64 target
10:42 not_gerd validation branch is ready for review, I believe
10:43 jnthn FROGGS: Hm, I don't have anything that looks like that... There are typically x86 and x64 shells with some stuff set up directly...
10:43 not_gerd error messages need to be made more helpful and we might decide to get rid of the flexibility pf validate_block/validate_sequence
10:44 not_gerd but selftest validates correctly and we should catch strictly more errors than master
10:47 * jnthn peers curiously at what the new bithacks.h is...
10:47 not_gerd well, right now just bit counting
10:47 not_gerd lunch&
10:48 jnthn +#error "Unsupported platform"
10:48 jnthn Might be a little nicer to say wat ain't supported, or even warn and have it provide a wild guess
10:49 jnthn Oh wow, windows. Great API. Count the darn bits!!
11:06 FROGGS jnthn: that is what I got, dont know how to produce english error messages https://gist.github.com/FR​OGGS/20fd21255473fe1c5684
11:08 jnthn OH NEINS ES IST IN KONFLIKT!
11:08 FROGGS *g*
11:08 FROGGS go ahead, make fun of me :P
11:08 jnthn Just your langauge :)
11:09 jnthn I'm wondering if you have a x64 environment, but if it's puling config from a 32-bit ActivePerl, or vice versa
11:11 FROGGS my perl is x64
11:11 not_gerd FROGGS: can you get at the actual compiler/linker invocation via --verbose-step=... ?
11:12 not_gerd jnthn: just be happy that I did not go with the windows API that supports more than 64 logical CPUs: http://msdn.microsoft.com/en-us/library/w​indows/desktop/ms683194%28v=vs.85%29.aspx
11:12 not_gerd if someone wants it, they can implement it themselves ;)
11:13 FROGGS not_gerd: you mean this?  https://gist.github.com/FR​OGGS/20fd21255473fe1c5684
11:13 jnthn whoa :)
11:15 not_gerd FROGGS: I missed the link line in that gist
11:15 not_gerd but yes, that is waht I wanted to look at
11:20 not_gerd FROGGS: does `echo %CURRENT_CPU% %TARGET_CPU%` tell you anything interesting?
11:21 FROGGS C:\MoarVM\nqp-cc\nqp\parrot>echo %CURRENT_CPU% %TARGET_CPU%
11:21 FROGGS %CURRENT_CPU% %TARGET_CPU%
11:22 FROGGS PROCESSOR_ARCHITECTURE=AMD64
11:22 FROGGS PROCESSOR_IDENTIFIER=Intel64 Family 6 Model 58 Stepping 9, GenuineIntel
11:22 FROGGS this is from my env
11:22 FROGGS CURRENT_CPU etc is not set
11:24 not_gerd the Windows SDK shell sets those
11:25 FROGGS I use the developer command prompt...
11:26 not_gerd do you have a LIBPATH set?
11:26 FROGGS LIB is set
11:27 FROGGS ohh, and libpath too, yeah
12:21 not_gerd o/
12:21 not_gerd left #moarvm
12:30 FROGGS damn, vs2010 doesnt work for me either :/
12:30 FROGGS seems like it is the same issue
12:33 jnthn FROGGS: Did yo check to see if your activeperl is 32 or 64? :)
12:33 FROGGS jnthn: it is still 64 bits, I counted them all you see :o)
12:36 FROGGS it is not that urgent though, I just wanted to have another test platform
12:48 FROGGS gah, that windows xp is screwed up too
12:48 FROGGS the cmd window closes when I execute vcvars32
12:50 JimmyZ what's your error?
12:52 FROGGS no error, the window just closes
12:53 FROGGS it did that too when I executed `git` instead of `git.exe`, re-installing git helped here
13:33 jnthn beer &
14:43 FROGGS meh
14:43 * FROGGS has no beer at home
14:54 tadzik heh. I wanted to say "what, this early in the morning?" but it's almost 5 PM
15:30 FROGGS if somebody with win7 could paste the out of the following, it would be super awesome:  perl.exe Configure.pl --optimize --verbose --prefix="C:/MoarVM/nqp-cc/nqp/install"
15:31 FROGGS (possibly agjust the prefix)
15:47 diakopter FROGGS: well, I have win8
15:48 FROGGS :/
15:48 diakopter but it shouldn't be any different for this
15:48 FROGGS but I think I know a workaround now
15:48 diakopter FROGGS: why you running vcvars32
15:48 FROGGS well, would be nice if you post it
15:49 diakopter this is what I run:
15:49 diakopter @"c:\Program Files (x86)\Microsoft Visual Studio 10.0\VC\vcvarsall" amd64
15:49 FROGGS because cmd + vcvars32 is the same as the vc2010 developer console
15:49 diakopter no
15:49 FROGGS hmmm
15:49 diakopter ^ that is
15:49 FROGGS my dev console just calls vcvars32
15:49 diakopter try that
15:49 FROGGS I'll try that
15:50 FROGGS C:\Users\leicht>@"c:\Program Files (x86)\Microsoft Visual Studio 10.0\VC\vcvarsa
15:50 FROGGS ll" amd64
15:50 FROGGS The specified configuration type is missing.  The tools for the
15:50 FROGGS configuration might not be installed.
15:50 FROGGS :/
15:51 diakopter hm
15:51 diakopter maybe you didn't install the 64 bit compiler
15:51 diakopter which would be weird, if you're on 64-bit
15:51 FROGGS k, I re-install it
15:53 diakopter this is my path
15:53 diakopter c:\Program Files (x86)\Microsoft Visual Studio 10.0\VC\BIN\amd64;C:\Windows\Microsoft.N​ET\Framework64\v4.0.30319;C:\Windows\Mic​rosoft.NET\Framework64\v3.5;c:\Program Files (x86)\Microsoft Visual Studio 10.0\VC\VCPackages;C:\Program Files (x86)\Microsoft Visual Studio 10.0\Common7\IDE;C:\Program Files (x86)\Microsoft Visual Studio 10.0\Common7\Tools;C:\Program Files (x86)\HTML Help Workshop;C:\Program Files (x86)\Microsoft Visual Studio ...
15:53 diakopter ... 10.0\Team Tools\Performance Tools\x64;C:\Program Files (x86)\Microsoft Visual Studio 10.0\Team Tools\Performance Tools;C:\Program Files (x86)\Microsoft SDKs\Windows\v7.0A\bin\NETFX 4.0 Tools\x64;C:\Program Files (x86)\Microsoft SDKs\Windows\v7.0A\bin\x64;C:\Program Files (x86)\Microsoft SDKs\Windows\v7.0A\bin;c:\users\mwilson\​src\moarvm\nqp-cc\install\bin;C:\Program ...
15:53 diakopter ... Files\Java\jdk1.7.0_25\bin;C:\Windows\​system32;C:\Windows;C:\Perl64\bin;C:\P​erl64\site\bin;c:\Users\mwilson\src\ic​u\bin;c:\Users\mwilson\bin;C:\Program Files (x86)\Git\cmd;C:\Python27;C:\Users\mwilson\src\gyp
15:54 diakopter (the result of running that command)
15:54 FROGGS looks like I am already missing the first directory, there is no \amd64 on my box
15:56 diakopter FROGGS: just download the vs 2010 premium iso+sp1 from me
15:56 diakopter Express is .... express.
15:57 FROGGS from you?
15:57 diakopter yah
15:58 diakopter privmsg
16:50 ingy *moarning*
16:56 BinGOs PATHs like that make baby Jesus cry.
16:57 BinGOs or is that nmake baby Jesus cry
17:31 ingy BinGOs++
18:16 FROGGS joined #moarvm
18:57 diakopter not_gerd++ core counter
19:06 dalek MoarVM/gcorch: b598b9e | diakopter++ | src/gc/ (4 files):
19:06 dalek MoarVM/gcorch: wip2
19:06 dalek MoarVM/gcorch: review: https://github.com/MoarVM/MoarVM/commit/b598b9e61a
19:17 FROGGS is it known that I dont see an output of say() under windows?
19:17 FROGGS C:\MoarVM\nqp-cc>..\moarvm.exe nqp.moarvm -e "say(42)"
19:17 FROGGS C:\MoarVM\nqp-cc>
19:26 diakopter echo %ERRORLEVEL%
19:26 diakopter FROGGS: ^
19:26 diakopter after that command
19:30 FROGGS C:\MoarVM\nqp-cc>..\moarvm.exe nqp.moarvm -e "say(42)"
19:30 FROGGS C:\MoarVM\nqp-cc>echo %ERRORLEVEL%
19:30 FROGGS -1073741819
19:30 diakopter crashed
19:30 diakopter FROGGS: which branch
19:31 FROGGS I compile mvm again
19:31 FROGGS I think I compiled it without the windows sdk, so maybe as 32bit
19:34 diakopter oh
19:34 diakopter well I need to fix that today
19:43 FROGGS k, it works now
19:45 dalek MoarVM/gcorch: 4a05985 | diakopter++ | src/gc/collect.c:
19:45 dalek MoarVM/gcorch: macroize something in collect.c
19:45 dalek MoarVM/gcorch: review: https://github.com/MoarVM/MoarVM/commit/4a05985f37
20:54 foo_bar_baz joined #moarvm
21:03 donaldh joined #moarvm
21:12 dalek MoarVM/gcorch: 4d81722 | diakopter++ | src/gc/collect.c:
21:12 dalek MoarVM/gcorch: ws
21:12 dalek MoarVM/gcorch: review: https://github.com/MoarVM/MoarVM/commit/4d81722cb6
21:40 Ulti joined #moarvm
21:45 FROGGS what I love about MoarVM too is that it behaves on Windows and linux identical
21:45 diakopter heh not always
21:45 diakopter configure/build, yes is awesome
21:46 FROGGS the selftest is identical, and that is worth much
21:46 FROGGS and yes, the build system seems pretty reliable
22:58 jnthn STOUT!!!
22:58 jnthn I mean, uh, evening :)
22:59 diakopter haha
22:59 diakopter :)
22:59 jnthn :)
23:00 * jnthn had a nice relaxing evening with $dayjob guy :)
23:00 dalek MoarVM: a9d954c | diakopter++ | build/Makefile.in:
23:00 dalek MoarVM: more consistent build output
23:00 dalek MoarVM: review: https://github.com/MoarVM/MoarVM/commit/a9d954c3bf
23:00 jnthn Our C++ teacher, actually :)
23:02 jnthn diakopter: Consistent as in...? :)
23:02 diakopter talking about the action instead of echoing the command like everything else
23:04 jnthn ah, k
23:04 jnthn I'm not sure what to think about that. It's less output, but if something fails..
23:08 diakopter you ok'd it before; recall there's a noisy flag
23:09 diakopter nmake realclean ; nmake NOISY=1
23:09 jnthn Just 'cus I OK something does not mean it's what I'd personally do. :)
23:10 diakopter it doesn't hide stderr output
23:11 diakopter .ask not_gerd what's the difference between typing nmake moarvm.dll and doing Configure.pl --shared
23:11 yoleaux diakopter: I'll pass your message to not_gerd.
23:12 jnthn In general, I'll only whine about or revert something if I think it's architecturally harmful. The rest, I try to tolerate. :)))
23:13 diakopter I'm un-architecturally harmful-izing the gcorch changes I made...
23:13 diakopter (speaking of..)
23:13 jnthn I'll read 'em tomorrow, probably :)
23:14 diakopter .. if I finish by then
23:14 jnthn (Got a 3 hour train trip to Goeteborg...)
23:14 jnthn OK :)
23:14 diakopter it led to shaving another ball of wax
23:14 diakopter er.
23:14 jnthn jnthn++ didn't order a train until 15:20, which allows lots of sleep :)
23:15 diakopter hey that's 55 min ago; you're late
23:15 jnthn diakopter: The GC orch code felt complex to me. If the overall effect is to make things clearer/simpler, it's good. :)
23:15 jnthn Swedish time, "tomorrow" :P
23:15 diakopter yes, too complex for correctness, too
23:15 jnthn Well, simple designs are easier to decide the correctness of.
23:16 jnthn A bunch of the stuff I spend my time teaching people is to be skeptical of their first idea of how to solve anything...
23:17 diakopter jnthn: btw if you and masak's software architecture class is any good, you can probably write a best-selling book on it.. there really aren't any good ones on that topic from what I can find
23:18 diakopter there are tons of books on "OO Design" and "Software Engineering" but none really focused on architeture I think.
23:19 diakopter well... or
23:19 jnthn It's hard to define "architecture"
23:20 jnthn I mean, you can ask 10 people and you'll get 10 different definitions.
23:20 jnthn In a MoarVM context, architecture is "stuff you commit that makes jnthn whine loudly", for example.

| Channels | #moarvm index | Today | | Search | Google Search | Plain-Text | summary