| diakopter | tadzik: so sleep indeed. | 00:08 | |
| lue | there's no way to specify multiple directories for --libpath, is there? I'm trying to make the perl6-m script work outside of rakudo :) | 00:17 | |
|
00:17
cognominal joined
00:35
colomon joined
01:00
grondilu joined
01:10
grondilu left
01:33
seaker joined
02:03
jnap joined
02:25
eternaleye joined
03:02
jnap joined
05:14
eternaleye joined
05:21
jnap joined
05:41
cognominal joined
|
|||
| diakopter realizes that type signatures are actually a type of grammar | 05:44 | ||
| .. and so are dependency statements | 05:56 | ||
| .. as is anything that requires any kind of logical resolution | 06:04 | ||
|
06:39
krunen joined
07:58
FROGGS joined
08:23
jnap joined
10:24
jnap joined
10:36
tgt joined
12:05
cognominal joined
13:13
lizmat joined
14:04
jnap joined
14:05
tgt joined
16:00
jnap joined
16:20
FROGGS joined
|
|||
| nwc10 | jnthn: I've found the cause of the GC garbage bug | 16:31 | |
| unconditionally pushing MVM args to the temp roots. Which gets all messy if they are optional named arguments that don't exist | |||
| lizmat | nwc10++ | 16:47 | |
| I'm pretty sure jnthn will be very glad when he sees this | 16:48 | ||
| TimToady | unless he sez sump'n like: but they're all initialized to something innocuous that could not be mistaken for a pointer | 16:49 | |
|
16:51
eternaleye joined
|
|||
| diakopter | .oO( doesn't Nick have a commit flag? ) |
17:18 | |
|
17:24
rurban joined
|
|||
| diakopter | er, push flag. | 17:30 | |
| TimToady | nwc10++ is not a pushy guy | 17:44 | |
|
18:36
ssutch joined
|
|||
| diakopter | nwc10: I can't find which branch you're on | 18:49 | |
| dalek | arVM: c506bb3 | diakopter++ | src/core/args.c: pushing optional arguments onto the temps stack if they are missing would be a bad thing (TM) nwc10++ |
18:53 | |
| diakopter | nm | ||
| TimToady: does yours make test now? | 18:54 | ||
| with --optimize, I mean | 18:57 | ||
| moritz: (or anyone) why hasn't moar-support been merged | 18:58 | ||
|
19:10
FROGGS[mobile] joined
|
|||
| FROGGS[mobile] | diakopter: I dunno, I asked jnthn++ but got no answer | 19:19 | |
| moritz has no idea | 19:23 | ||
| maybe because the setting compilation still occasiontally segfaults? | 19:24 | ||
| also, would somebody please give nwc10++ a commit bit? :-) | |||
| (for moar, that is) | |||
| timotimo | something about not wanting to commit to a serialisation version or something? | 19:27 | |
| FROGGS[mobile] | who can do that? | ||
| timotimo | we also still need to do the code that puts blobs of data into the moar bytecode files rather than base64 encoding them | 19:28 | |
| FROGGS[mobile] | timotimo: we dont have problem I trink | ||
| timotimo | what do you trink? :) | 19:29 | |
| FROGGS[mobile] | (the sc version thing) | ||
| ohh, I missed a "that" | |||
| timotimo | hm, ok | ||
| still, getting rid of the base64 stuff would probably be a little win for startup time :3 | 19:30 | ||
| FROGGS[mobile] | go ahead I'd say, but it sounded aß there is some design work needed | 19:31 | |
| as* | |||
|
19:33
eternaleye joined
|
|||
| timotimo | maybe it's easier than you all think :P | 19:33 | |
| (i don't actually think i can just do it) | |||
| FROGGS[mobile] | I guess we will need a MVMBlob to store binary stuff | 19:36 | |
| timotimo | Test.pm still segfaults even with the fix diakopter just committed for nwc | 19:39 | |
| diakopter | :( | 20:08 | |
| FROGGS | not for me | ||
| make test in rakudo is clean now (for the first time) | |||
| diakopter | with --optimize? | ||
| FROGGS | with --optimize=1 | 20:09 | |
| diakopter now reads #perl6 :) | |||
| I'll try to reproduce timotimo's err | 20:10 | ||
| timotimo: what's your system (so I can download the same distro/kernel) | 20:11 | ||
| diakopter is feeling very ambitious today, being soon surrounded by 100 hacktivists | 20:12 | ||
| FROGGS | *g* | ||
| timotimo | diakopter: it's a fedora 20 | 20:13 | |
| diakopter | #g0v.tw on freenode in case you're curious | ||
| timotimo | er, wrong | ||
| that's remote | |||
| it's a linux mint | |||
| diakopter | :) | ||
| 32/64? version? | |||
| timotimo | Linux kischde 3.8.0-33-generic #48-Ubuntu SMP Wed Oct 23 09:16:58 UTC 2013 x86_64 x86_64 x86_64 GNU/Linux | ||
| 64 | |||
| diakopter | clang or gcc? | ||
| timotimo | gcc afaict | ||
| gcc (Ubuntu/Linaro 4.7.3-1ubuntu1) 4.7.3 | |||
| diakopter | timotimo: which should I download? :√ community.linuxmint.com/iso | 20:16 | |
| great, now I have to choose between Cinnamon desktop and MATE desktop, neither of which I've heard of :Ω | 20:19 | ||
| timotimo | that's a really good question. i never saw this page before | ||
| if you have a beefy graphics card, you can safely go cinnamon | |||
| diakopter | ok :∂ | ||
| timotimo | mate is a gnome2 fork | ||
| :Γ | 20:20 | ||
| diakopter | gee thanks safari for displaying the utf8 of the .iso in my browser window. just what I wanted. :) | ||
| timotimo | :D | ||
|
20:31
colomon joined
|
|||
| diakopter | well I'll get mint 16; hopefully it still has the problem :) | 20:39 | |
| TimToady uses cinnamon, but hasn't tried to compile on mint 16 yet | 20:45 | ||
| TimToady is watching the perl6-m compile go past, and wondering why --encoding=utf8 is not the default | 20:55 | ||
| perl6-m compile successful on Mint 15 | 21:00 | ||
| (64-bit, cinnamon) | 21:03 | ||
| diakopter | is optimize=1 the default nowadays? | ||
| TimToady | I believe it used 3 | ||
| yes, 3 | 21:05 | ||
| diakopter | me 10 years from now: "how many internets does your local internet backup archive contain?" "all of them except the quintillion most recent ones" | 21:08 | |
| TimToady | earlier success was on my 6-core at home; it failed to compile Test on my laptop though | ||
| diakopter hasn't built nqp/rakudo in some .. months | 21:09 | ||
| TimToady | trying a reconfig | ||
| diakopter | let alone on moar. so I'm curious to see how fast. | ||
| FROGGS | p: sub caller($o = 0) { try die; Backtrace.new[$o + 2] }; say caller(0) # I love it | 21:10 | |
| camelia | rakudo-parrot 413650: OUTPUT«Backtrace::Frame.new(file => "/tmp/Fq0i3AufIY", line => 1, code => sub caller($o = { ... }) { ... }, subname => "caller")» | ||
| FROGGS | +3 actually... | 21:11 | |
| ww :o) | 21:12 | ||
| diakopter | mint boots absurdly quickly | 21:14 | |
| TimToady | security updates from ubuntu are absurdly slow though, last I knew | 21:15 | |
| diakopter | "a new version of the update manager is available" | 21:16 | |
| ok. | |||
| "enter your password to start the update manager" | 21:17 | ||
| ok. | |||
| .. and now 172 updates of 233 MB. | 21:18 | ||
| yay | |||
| well it's install gcc 4.8, and timotimo has 4.7 | 21:19 | ||
| *installing | |||
| bah | |||
| TimToady | Test still fails on laptop, blowing away install and nqp now | 21:22 | |
| I wonder if it's just a function of less memory available on the laptop; thought jnthn++ said not a week or two ago | 21:23 | ||
| diakopter | url? | ||
| this has got to be the hundredth time I've spent several minutes waiting for libreoffice and thunderbird to download/install without ever intending to use them. | 21:25 | ||
| thousandth if you count updates, probably | 21:26 | ||
| TimToady: what else did jnthn say about that | 21:27 | ||
| how could limited memory trigger the different behavior | |||
| TimToady | he said it couldn't, cause the allocation was fixed | 21:28 | |
| not based on how much memory you have | |||
| diakopter | yeah; currently it never frees any of the main heaps until the very end | 21:29 | |
| however, maybe therein lies the clue. if malloc changes its allocation strategy based on constrained memory, perhaps we are hanging onto some freed malloc'd pointer. | 21:33 | ||
| TimToady | I don't think malloc cares about that | 21:34 | |
| diakopter | but there are lots of mallocs | ||
| maybe some reduce the break | |||
| or whatever it's called | |||
| TimToady | I've never met a malloc that cares about that | ||
| but if it reduces the break, it would be unlikely to do that based on total system memory, methinks | 21:35 | ||
| usually it's just, "oh, the last thing before the break went away, so I can move the break back" | 21:36 | ||
| malloc isn't allows to move things around... | |||
| *wed | |||
| diakopter | but maybe if there's less available, it tries harder to avoid fragment | 21:37 | |
| TimToady | maybe, but I've never met that malloc | 21:38 | |
| diakopter | years later, git still depends on perl for a trivial-ish function. oddity of oddities. | 21:40 | |
.oO( guardian angel.. coattails.. ) |
|||
| mwilson@host22 ~/src/rakudo $ perl Configure.pl --gen-moar | 21:48 | ||
| Use of uninitialized value $default_backend in substr at Configure.pl line 121. | |||
|
21:50
jnap joined
|
|||
| FROGGS | FIX IT"!!1 | 21:50 | |
| diakopter | no rakudo commit bit :) | 21:51 | |
| FROGGS | haha | ||
| diakopter | (and forced ignorance to submit a pull request) | ||
| FROGGS | um, you are not even joking O.o | 21:53 | |
| diakopter | well.. | ||
| while also poking fun/self-deprecating accurately | |||
| mint gave me gcc without giving me libc-dev | 21:54 | ||
| *headdesk* | |||
| TimToady | how's it supposed to guess you're a dev? | 21:56 | |
| diakopter | :) | ||
| TimToady | most people aren't, or shouldn't... | ||
| diakopter | well that was quick. | ||
| FROGGS | and especially a libc-dev | ||
| diakopter | /home/mwilson/src/rakudo/install/bin/moar --libpath=src/vm/moar/stage0 src/vm/moar/stage0/nqp.moarvm --bootstrap --setting=NULL --no-regex-lib --target=mbc \ --output=gen/moar/stage1/nqpmo.moarvm gen/moar/stage1/nqpmo.nqp | ||
| make: *** [gen/moar/stage1/nqpmo.moarvm] Segmentation fault | |||
| very first thing it tried to use moar for, it dieded | 21:57 | ||
| TimToady: well, -dev is the thing that contains the std .h | 21:59 | ||
| TimToady | my laptop still consistently dies on Test.pm, where my home computer worked. I wonder what's different... | ||
| diakopter | I can't build nqp at all | ||
| TimToady | fresh from git? | 22:00 | |
| diakopter | right | ||
| 'course, I didn't have libc much of the way through. | 22:01 | ||
| trying again. | |||
| TimToady | my nom-level configuration is: perl Configure.pl --backends='parrot,jvm,moar' --prefix='/home/larry/nom/install' --gen-parrot --gen-nqp=master --gen-moar=master | ||
| note the explicit =masters | 22:02 | ||
| diakopter | I wish searching for a commit hash on google would link directly to the proper commit on github | ||
| hm | |||
| oh ok. | 22:03 | ||
| well I wonder why the moar-support doesn't do that :) | |||
| ah, that's better. | |||
| TimToady | it also required the absolute path to install, relative didn't work, and maybe still doesn't | 22:04 | |
| diakopter | I've never seen nqp build that quickly | 22:05 | |
| TimToady | well, Configure often doesn't rebuild when you'd want it to, so sometimes you have to blow away nqp and install | 22:06 | |
| TimToady --> nap, try to get well | 22:08 | ||
| diakopter | Stage start : 0.000 | ||
| Stage parse : 106.056 | |||
| Stage syntaxcheck: 0.000 | |||
| Stage ast : 0.000 | |||
| Stage optimize : 12.209 | |||
| does that seem right? | |||
| [Coke] | sureish. | ||
| diakopter | oh | ||
| Stage mast : 81.834 | 22:09 | ||
| Stage mbc : 0.924 | |||
| Segmentation fault | |||
| make: *** [CORE.setting.moarvm] Error 139 | |||
| *cry* | |||
| FROGGS | make install again | 22:13 | |
| sometimes it fails after building the setting | |||
| diakopter | gah. | ||
| perl -MExtUtils::Command -e cp /home/mwilson/install/bin/perl6-m /home/mwilson/install/bin/perl6 | 22:14 | ||
| perl -MExtUtils::Command -e chmod 755 /home/mwilson/install/bin/perl6 | |||
| Cannot chmod 755 /home/mwilson/install/bin/perl6:No such file or directory at /usr/share/perl/5.14/ExtUtils/Command.pm line 274. | |||
| so.. how could make install think that there's a perl6-m when there's not | |||
| ah, there's a bug | 22:17 | ||
| m-all: $(PERL6_MOAR) $(SETTING_MOAR) $(M_RUNNER) lib/Test.moarvm blib/lib.moarvm blib/Pod/To/Text.moarvm | 22:18 | ||
| $(M_RUNNER) needs to have its path | |||
| er | 22:20 | ||
| mwilson@host22 ~/src/rakudo $ ./perl6-m -e 'hi' | 22:23 | ||
| ===SORRY!=== | |||
| p6sort NYI | |||
| moar: 3rdparty/libuv/src/unix/loop.c:150: uv__loop_delete: Assertion `!((((*(&(loop)->active_reqs))[0]) == (&(loop)->active_reqs)) == 0)' failed. | |||
| Aborted | |||
| FROGGS: ^ what does this mean | |||
| [Coke] | diakopter: try with --ll-exception | 22:24 | |
| (given that 'hi' isn't valid, might be dying on error reporting) | |||
| diakopter | yeah | ||
| what does that do | |||
| (what does the ll stand for) | 22:25 | ||
| low-level I'll assume | |||
| [Coke] | low level | ||
| FROGGS | diakopter: that libuv assertion always pops up if you don't exit cleanly | 22:27 | |
| diakopter | gah :( | 22:28 | |
| FROGGS | ./perl6-m -e 'say "hi"' should work though | 22:30 | |
| diakopter | yeah, does | 22:32 | |
| FROGGS | what optimize flags does moar got? | 22:38 | |
| diakopter | 3 | ||
| FROGGS | interesting, I guess it still fails on my box | ||
| diakopter | I thought moar had a --show-config | ||
| r: say "","","" | 22:41 | ||
| camelia | rakudo-parrot 2b027b, rakudo-jvm 2b027b: OUTPUT«» | ||
| diakopter | r: say ".",".","." | ||
| camelia | rakudo-parrot 2b027b, rakudo-jvm 2b027b: OUTPUT«...» | ||
| diakopter | oh. | ||
| FROGGS | nqp has | 22:44 | |