Perl 6 - the future is here, just unevenly distributed

IRC log for #moarvm, 2014-11-29

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

All times shown according to UTC.

Time Nick Message
00:50 kjs_ joined #moarvm
06:39 lue joined #moarvm
07:21 dalek MoarVM: dfdef4c | TimToady++ | src/6model/reprs/NFA.c:
07:21 dalek MoarVM: replace quicksort with insertion sort
07:21 dalek MoarVM:
07:21 dalek MoarVM: N is generally small, so a simple insertion sort is more appropriate,
07:21 dalek MoarVM: especially when moved to the point we already have the fates in the cache.
07:21 dalek MoarVM: review: https://github.com/MoarVM/MoarVM/commit/dfdef4c92c
08:42 lizmat joined #moarvm
09:54 woolfy joined #moarvm
11:49 JimmyZ_ joined #moarvm
12:12 jnthn wtf was I thinking in frame.c:555...
12:13 jnthn That's rooting osmething that's not even a GC-able thing, it's a register frame array...
12:14 tadzik joined #moarvm
12:15 synopsebot joined #moarvm
12:19 JimmyZ_ Oh, How did it work :)
12:20 jnthn By getting hit rarely, I guess...
12:22 lizmat reason for OS X flakiness ?
12:22 JimmyZ_ so only OS X hit it ? ;)
12:22 jnthn lizmat: Well, I took a spectest that was sometimes failing, did a bunch of GC setting tweaks, and got same test to explode due to this reason under the debugger.
12:23 jnthn lizmat: I can confirm that particular issue I found on OSX goes away. I now need a fresh normal build and to do a clean build/spectest.
12:24 * lizmat is looking forward to the results
12:25 jnthn If clang ever finishes compiling interp.c, we'll find out... :P
12:26 lizmat brb&
12:35 jnthn Well, further improvement.
12:36 jnthn All passed except an explosion in t/spec/S04-statements/terminator.t
12:38 JimmyZ_ That's very nice
12:40 lizmat jnthn++
12:42 jnthn That failure is, unfortunately, still indicative of a bug somewhere.
12:43 jnthn But this is by far the cleanest I've seen OSX since I started looking into the issues.
12:50 lizmat yeah, reason enough for a bump ?
12:54 jnthn lizmat: yes, let's
12:54 jnthn It'll give us some more feedback
12:54 lizmat ok, will do
12:55 * timotimo is pleased
13:40 kjs_ joined #moarvm
14:32 kjs_ joined #moarvm
14:36 dalek joined #moarvm
15:04 brrt joined #moarvm
15:04 brrt \o
15:06 JimmyZ o/
15:13 timotimo |o
15:16 brrt oh good news... jnthn++
15:39 kjs_ joined #moarvm
15:44 zakharyas joined #moarvm
16:20 zakharyas joined #moarvm
16:36 lizmat well... the feedback from me so far, is that I still see random testfiles fail during spectest
16:42 TimToady maybe it shoulda been named osx canary
16:45 jnthn lizmat: :/
16:49 jnthn Guess we'll have to see what improvement the [Coke]++ daily runs show up
16:54 jnthn stouting &
17:52 FROGGS_ joined #moarvm
18:15 woolfy left #moarvm
18:36 lizmat joined #moarvm
18:47 ilbot3 joined #moarvm
18:47 Topic for #moarvm is now https://github.com/moarvm/moarvm | IRC logs at  http://irclog.perlgeek.de/moarvm/today
19:57 kjs_ joined #moarvm
19:58 kjs_ joined #moarvm
20:15 hoelzro FROGGS_: have you had a chance to look at that precomp bug?
20:16 FROGGS_ hoelzro: no, $dayjob keeps me busy :o(
20:16 FROGGS_ (also right now)
20:16 hoelzro ah ha =/
20:17 hoelzro I looked at it a bit, and I *thought* I was on to something, but jnthn assured me that's not the issue
20:18 FROGGS_ ahh, k
20:18 hoelzro I think my problem is that my knowledge of how precomp works is very limited
20:18 FROGGS_ I should ask what you were after then before I start getting into it
20:19 FROGGS_ I know some details but my problem is that my brane is not smart enough to keep all details in memory at the same time :o)
20:20 hoelzro well, I thought it was because GLOBALish isn't passed in here: https://github.com/rakudo/rakudo/blob/nom/src/Perl6/World.nqp#L361
20:20 hoelzro but jnthn said that GLOBALish is serialized with all of the correct linkage, so that's not it
20:20 hoelzro that's as far as I got
20:21 dalek MoarVM: c6354db | TimToady++ | src/6model/reprs/NFA.c:
20:21 dalek MoarVM: more cleanups; carp on wrong # of fates from nqp
20:21 dalek MoarVM: review: https://github.com/MoarVM/MoarVM/commit/c6354db280
21:54 timotimo jnthn: do you think recording where the code currently was when a GC got triggered would be interesting?
21:57 timotimo jnthn: when forcing the GC to run, the profiler output for GC runs is wrong, it shows a green bar all the way to the right, even if the three parts don't add to the same amount
22:02 lizmat joined #moarvm
22:09 timotimo i'm not sure if i should make the progress bar shorter or add a "wasted" section
22:36 [Coke] joined #moarvm
22:38 * [Coke] finally kicks off today's runs.
22:44 lizmat [Coke]++
23:11 colomon joined #moarvm
23:48 timotimo 00062   const_s      loc_1_str,
23:48 timotimo 'CwAAAEAAAAAFAAAAaAAAABwAAAC4AQAAZl4AAHICAACGhQAAGdAAACkAAADx0wAAFQAAAEHVAAAB2wAAAAAAAAUAAAAGAAAAcwAAAHQAAADbAAAA3AAAAN8AAADgAAAAKgEAACsBAAABAAAAAAAAAFAAAAABAAAAUAAAAKAAAAADAAAAoAAAAEQHAAB1AAAARAcAAJ4TAADYAAAAFRQAAEMbAAADAAAAXRsAAB0iAAABAAAAHSIAAG0iAAB1AAAAbSIAALkrAAB1AAAA9CwAAEA2AAB1AAAAezcAAMdAAAB1AAAAAkIAAE5LAAB1AAAAiUwAANVVAAABAAAAEFcAAFhXAAABAAAAWFcAAPpXAAABAAAA+lcAAEJYAAABAAAAQlgAAKhYAAABAAAAqFgAAPBYAAA
23:48 timotimo BAAAA8FgAAFZZAAABAAAAVlkAAJ5ZAAABAAAAnlkAAARaAAABAAAABFoAAExaAAABAAAATFoAALJaAAABAAAAsloAAPpaAAABAAAA+loAAGBbAAABAAAAYFsAAKhbAAABAAAAqFsAAA5cAAABAAAADlwAAF5cAAABAAAAXlwAAK5cAAAAAAAApAAAAAAAAAAAAAAAAgAAAAAApQAAAAEAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAgAAAAAAAACmAAAAAAAAAAEAAAACAAAAAACnAAAAAQAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAACAAAAAAAAAKgAAAAAAAAACgAAAAI
23:48 timotimo AAAAAAKkAAAAKAG0AAAAEAAAAAgABAAAAX8EAAAcAAAACAAEAAADCBgAACAAAAAIAAQAAAMoGAAAJAAAAAgABAAAA4gYAAAoAAAACAAEAAADqBgAACwAAAAIAAQAAAPoGAAAMAAAAAgABAAAABAcAAA0AAAACAAEAAAATBw
23:48 timotimo how come we're still putting serialized blobs in base64?

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