Perl 6 - the future is here, just unevenly distributed

IRC log for #moarvm, 2016-03-25

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

All times shown according to UTC.

Time Nick Message
01:07 timotimo my headache is limiting how well i can work here ... it took me *long* to notice i hadn't actually recompiled nqp after the patch to fix compunit pushing
01:10 diakopter *comiserate*
01:16 timotimo i don't know what that means :(
01:53 timotimo when i'm feeling better i'll probably introduce graphviz support into the heapanalyzer
02:02 timotimo why is graphviz.org down :\
02:03 timotimo seems like connectivity in general is kind of b0rked from here right now :|
02:04 timotimo google cache <3
02:05 timotimo it doesn't seem like graphviz has anything in its -Tx11 that'd react to clicks on specific nodes :\
02:06 timotimo otherwise it'd be totally feasible to use it as a front-end gui like thing and adding expand/collapse/path buttons to every node in the graph
02:06 timotimo together with its "automatically refresh when the file changes" behavior
02:13 colomon joined #moarvm
02:13 geekosaur I recall seeing that the unstable graphviz had a mode with more clickability... but the whole thing was still limited. Ended up not being my problem any more, and I think $coworker switched it to something else
02:15 timotimo oh, huh
03:05 diakopter did you look up comiserate :D
03:19 dalek joined #moarvm
05:14 vendethiel joined #moarvm
05:50 dalek joined #moarvm
06:07 dalek joined #moarvm
06:08 mojca joined #moarvm
06:55 timotimo yeah, it doesn't exist in my favorite dictionary
06:56 timotimo it does find it once you add the second m to it
06:56 timotimo and now that i see the translation, it's quite obvious that it means that, given its composition
06:56 timotimo anyway, thanks :)
07:27 mojca joined #moarvm
07:40 domidumont joined #moarvm
07:45 domidumont joined #moarvm
08:18 dalek joined #moarvm
08:29 dalek joined #moarvm
10:08 FROGGS joined #moarvm
10:45 vendethiel joined #moarvm
11:29 sivoais joined #moarvm
12:01 vendethiel joined #moarvm
12:33 vendethiel joined #moarvm
13:31 vendethiel joined #moarvm
14:59 vendethiel joined #moarvm
16:38 vendethiel joined #moarvm
17:03 diakopter ohh
17:36 dalek Heuristic branch merge: pushed 130 commits to MoarVM/cache_sc_idx by diakopter
17:46 cognominal joined #moarvm
17:52 colomon joined #moarvm
18:09 domidumont joined #moarvm
18:20 vendethiel joined #moarvm
18:41 colomon joined #moarvm
18:56 vendethiel joined #moarvm
19:03 diakopter aww, the cache_sc_idx branch only improves setting compile time by 7% now
19:04 diakopter ISTR it was higher a bit ago
19:12 diakopter does anyone get a clean rakudo spectest these days from moar master?
19:25 timotimo hm, hat does perl6-roast-data say?
19:25 diakopter dunno.
19:25 diakopter but now (with strawberry toolchain), spectesting is hanging in S09-typed-arrays
19:25 diakopter (and master branch)
19:25 diakopter (not cache_sc_idx)
19:26 diakopter anyone else build with strawberry?
19:28 timotimo huh, in the typed arrays test?
19:29 diakopter maybe this is interesting: all the stacks are hung in acquiring a spinlock in the kernel, called from MVM_gc_write_barrier_hit, called from MVM_crash_on_error
19:30 vendethiel joined #moarvm
19:30 diakopter so I guess the spectest harness is having trouble detecting the hangs or killing them
19:31 diakopter well I guess they wouldn't respond to signals if they're stuck in the kernel
19:33 timotimo huh, it's supposed to crash, but it's trying to acquire some lock?
19:33 diakopter yah
19:35 timotimo that's definitely not so good :)
19:35 diakopter whatevs, I'll just skip the strawberry toolchain and go directly to VS 2015 update 2
19:40 diakopter after all, if it doesn't work in master, it's not really a supported platform, heh
19:44 diakopter boy NQP really builds fast these days
19:47 timotimo aye
19:48 timotimo i get an almost clean spectest run
19:48 timotimo t/spec/S32-exceptions/misc.rakudo.moar  -  non-zero exit code
19:48 timotimo t/spec/S32-num/power.rakudo.moar  -  passes a bunch of TODOs
19:48 diakopter well I already abandoned strawberry entirely
19:49 diakopter who cares about gcc on windows, anyway
19:49 diakopter .seen lizmat
19:53 diakopter rakudo-moar on Visual Studio builds 20% faster than on gcc anyway
19:57 diakopter aaaaaaaand we're hung in S09-typed-arrays again
19:58 diakopter weird
19:59 diakopter maybe nqp needs rebootstrapped and hasn't been bumped?
20:06 diakopter strange, entirely different toolchain and PATH, entirely different checkout
20:06 * diakopter looks around for the JNTHN
20:20 timotimo you only need to rebootstrap nqp if you're on the branch that does the rebootstrap for you
20:20 diakopter oh
20:20 diakopter well I'm surprised the S09-typed-arrays tests seem not to work on Windows at all
20:20 timotimo that surprises me, too
20:21 diakopter I mean, I guess it could've been a recent commit that wasn't spectested on windows
20:21 timotimo could be, yeah, but jnthn does that all the time
20:22 diakopter I mean recent like past couple days :D
20:22 diakopter oh well
20:32 kjs_ joined #moarvm
20:58 kjs_ joined #moarvm
22:40 lizmat diakopter: do you have a gist of where it hangs ?
22:41 lizmat diakopter: there used to be a hang with .prepend: are you sure you're testing the latest ?
22:44 mojca joined #moarvm
22:47 diakopter yes, tips of everything
22:47 diakopter all I can see is the C stack trace there
22:47 lizmat is there a specific one that fails, or do they all fail ?
22:48 diakopter seems like all
22:48 diakopter I'm thinking it's gotta be a commit in the past few days (since jnthn spectested on windows)
22:49 diakopter since it fails the same way on both gcc/strawberry and msvc
22:52 lizmat diakopter: I would be very surprised to see OS dependent behaviour in that part of the code
22:52 lizmat it feels more to me that you're testing with an older rakudo, *with* the .prepend bug still in
22:52 lizmat because I fixed that, and *then* added tests for it
22:53 diakopter I did git clone
22:53 diakopter fresh for both gcc and msvc
22:53 lizmat make t/spec/S09-typed-arrays/*.t   # passes for me without any errors
22:53 lizmat so I'm not sure how I can help with this...
22:54 lizmat at least not without a gist
22:54 diakopter what kind of gist
22:55 lizmat perl6 t/spec/S09-typed-arrays/native-str.t
22:55 lizmat for instance
22:55 lizmat (assuming you've done a make install)
22:57 diakopter invoking perl6 didn't work
22:57 diakopter (can't find Test)
22:57 diakopter oh FFS
22:58 lizmat perl6 -Ilib t/spec/S09-typed-arrays/native-str.t
22:58 diakopter the problem is the \r in my path
22:58 diakopter ridiculous
22:59 timotimo oh lord
22:59 diakopter in the error for not finding Test
22:59 diakopter three of the paths lost the first part (C:\Users\mwilson\src\v\r)
22:59 lizmat I'm not sure I follow anymore
22:59 diakopter they instead start with akudo\install\share\perl6
23:00 diakopter somehow the \r is being interpolated
23:00 diakopter and maybe the \v also?
23:01 diakopter it says:
23:01 diakopter Could not find Test in:
23:01 diakopter C:\Users\mwilson\.perl6
23:01 lizmat could you try: perl6 -Ilib t/spec/S09-typed-arrays/native-str.t
23:01 lizmat ?
23:01 diakopter akudo\install\...
23:02 lizmat and maybe submit an error report about the error report being incorrect ?   :-)
23:02 diakopter I did
23:02 diakopter (just now when you suggested)
23:02 diakopter it succeeded (all passed)
23:02 lizmat *phew*  :-)
23:02 diakopter I don't think the error report is incorrevct
23:03 lizmat if it eats \r and \v's, it is, isn't it ?
23:03 diakopter it was wrong in my Makefile too
23:03 diakopter so it's brokener before the error message
23:03 diakopter I had to add the \v and \r and remove a line break from two spots in the Makefile too
23:03 lizmat it was my understanding that all moderner Windows understand / in paths
23:04 diakopter only from scripts, not from shell
23:04 lizmat isn't this from scripts ?
23:04 diakopter well, it works from shell I guess
23:04 diakopter but anyway, I didn't do anything special in Configure
23:04 diakopter so Configure spat it out wrong
23:04 lizmat so, my point being, why don't we just use forward slashes, even on Win ?
23:05 diakopter well it already normalizes to something
23:05 diakopter even if the user passes in paths with backslashes in configure options?
23:07 * lizmat is tired after cycling 64km through hilly terrain
23:07 lizmat so good night, #moarvm
23:07 diakopter o/
23:09 timotimo gnite liz
23:45 kjs_ joined #moarvm

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