| timotimo | the reverse-cont i set up with a watchpoint has been running for what feels like 4 hours now | 01:19 | |
| ok, looks like rr kind of died | 01:27 | ||
|
02:07
Kaiepi joined
02:40
FROGGS_ joined
02:56
ilbot3 joined
|
|||
| MasterDuke | reverse-cont? | 03:23 | |
|
03:40
unicodable6 joined
04:47
Ven`` joined
05:32
bisectable6 joined
06:46
greppable6 joined
07:16
domidumont joined
07:25
domidumont joined
07:31
robertle joined
07:59
robertle joined
08:16
lizmat joined
|
|||
| Geth | MoarVM: 67056c2041 | (Stefan Seifert)++ | src/spesh/optimize.c Fix optimization of zombie conditional causing segfault optimize_iffy may end up removing a conditional altogether if the tested value is known. In this case the call to decompose_object_conditional right after would process an already removed conditional instruction and insert a new instruction for processing the same input as the original conditional. However the original conditional may have been the only instruction keeping ... (5 more lines) |
08:51 | |
|
09:03
travis-ci joined
|
|||
| travis-ci | MoarVM build failed. Stefan Seifert 'Fix optimization of zombie conditional causing segfault | 09:03 | |
| travis-ci.org/MoarVM/MoarVM/builds/357720705 github.com/MoarVM/MoarVM/compare/a...056c20416a | |||
|
09:03
travis-ci left
|
|||
| nine | No idea what this ^^^ is about | 09:05 | |
|
11:01
committable6 joined
11:36
lizmat joined
11:48
domidumont joined
11:55
Ven`` joined
|
|||
| MasterDuke | nine: have you seen brrt's branch about fixing something in optimize_iffy? | 12:03 | |
| nine | no | 12:15 | |
| MasterDuke | nine: github.com/MoarVM/MoarVM/tree/spes...actor-iffy i think | 12:24 | |
| nine | MasterDuke: I think github.com/MoarVM/MoarVM/commit/87...eda7c88f78 would have fixed the issue, too | 13:12 | |
| timotimo | if i can get that error locally, i can figure out what part makes it go boom | 13:28 | |
| i'd really like travis to re-run the last command that failed with a few different flags, like MVM_SPESH_LOG, so that we can grab the logs from the artifacts part of the build | 13:30 | ||
|
13:51
AlexDaniel joined
14:37
Ven`` joined
14:38
committable6 joined
|
|||
| Geth | MoarVM: b5b69681f6 | (Stefan Seifert)++ | src/spesh/optimize.c Fix "'optimize_iffy' : not all control paths return a value" |
15:07 | |
|
16:03
bisectable6 joined
|
|||
| timotimo | fish: “python syncthing-gtk.py” terminated by signal SIGBUS (Misaligned address error) | 16:55 | |
| what have i done?! | |||
|
17:02
Ven`` joined
|
|||
| timotimo | why's my hard drive so full? oh, a 1.1 gig big heap snapshot, okay! | 17:18 | |
| so which of my konsole sessions is the one with 1.8 gigs of scrollback stored on my hard drive | 17:20 | ||
|
17:45
Ven`` joined
|
|||
| Geth | MoarVM/grisu: 3ad2183389 | (Zoffix Znet)++ | 4 files Stringify Num using Grisu3 algo - Makes Num stringification 2x faster (tested with rand.Str) - Fixes RT#127184 rt.perl.org/Ticket/Display.html?id=127184 - Fixes RT#124796 rt.perl.org/Ticket/Display.html?id=124796 - Fixes RT#132330 rt.perl.org/Ticket/Display.html?id=132330 (fixes Num.WHICH and problems with set()s mentioned in that ticket) ... (40 more lines) |
19:17 | |
| synopsebot | RT#127184 [open]: rt.perl.org/Ticket/Display.html?id=127184 [BUG] One digit short to correctly stringify a double | ||
| synopsebot | RT#124796 [new]: rt.perl.org/Ticket/Display.html?id=124796 Roast rakudo skip/todo test:./S32-num/stringify.t line:69 reason: 'big stuff' | ||
| RT#132330 [open]: rt.perl.org/Ticket/Display.html?id=132330 [MATH] Floating point anomalies | |||
| AlexDaniel | Zoffix++ | 19:24 | |
| dogbert11 is a bit worried about the NQP bump | 19:25 | ||
| AlexDaniel | dogbert11: ok, what's wrong with it? | 19:31 | |
| dogbert11 | same problem as before, when you bumped | ||
| dilbert@Linux-Mint18 ~/repos/rakudo $ ./perl6 t/02-rakudo/03-cmp-ok.t | |||
| ===SORRY!=== | |||
| Missing or wrong version of dependency 'gen/moar/Metamodel.nqp' (from '/home/dilbert/repos/rakudo/lib/Test.pm6 (Test)') | |||
| AlexDaniel | try nuking rakudobrew again? | 19:32 | |
| dogbert11 | I'm not using rakudobrew now :) | ||
| can you run that test with the latest version? | 19:33 | ||
| there are 3 possibilities: 1/ I'm a moron, 2/ the build process is broken, 3/one of the NQP commits is no good | 19:35 | ||
| AlexDaniel | dogbert11: travis did that without issues | 19:39 | |
| dogbert11 | so we're leaning towards alternative 1 then :) | 19:40 | |
| MasterDuke | both my laptop and desktop built nqp+rakudo fine after the bumps | ||
| dogbert11 | hmm, this is strange | ||
| dogbert11 runs another spectest after wiping t/spec/packages/.precomp | 19:45 | ||
| timotimo | zoffix++ good work! i saw the screenshot on twitter :D | 19:56 | |
| zoffix, if you want something purely for performance, this is an algo that we probably can just put into 3rdparty that does integers quickly: pvk.ca/Blog/2017/12/22/appnexus-co...rs-faster/ | 20:02 | ||
| dogbert11 | had to wipe out several .precomp directories in order to get everything to work properly | 20:39 | |
| timotimo | huh. we're supposed to not load precomp files that don't fit our stuff | ||
| did something go wrong in nine's reproducible builds patch? | |||
| Geth | MoarVM/grisu: 067c059410 | (Zoffix Znet)++ | 4 files Stringify Num using Grisu3 algo - Makes Num stringification 2x faster (tested with rand.Str) - Fixes RT#127184 rt.perl.org/Ticket/Display.html?id=127184 - Fixes RT#132330 rt.perl.org/Ticket/Display.html?id=132330 (fixes Num.WHICH and problems with set()s mentioned in that ticket) ... (39 more lines) |
20:40 | |
| synopsebot | RT#127184 [open]: rt.perl.org/Ticket/Display.html?id=127184 [BUG] One digit short to correctly stringify a double | ||
| synopsebot | RT#132330 [open]: rt.perl.org/Ticket/Display.html?id=132330 [MATH] Floating point anomalies | ||
| dogbert11 | timotimo: dunno, perhaps my setup is different since I'm the only one complaining | 20:48 | |
| Geth | MoarVM/grisu: 8841c4241b | (Zoffix Znet)++ | src/math/grisu.c Mod post-Grisu3 Num renderer… …to prefer normalized forms over the shortest possible ones. - Also, add a bit of fudging to maintain old format, such us collapsing into scientific form for `e > 14` and adding leading zero to 1-digit exponents. - Also, toss the assert()s #YOLO |
21:43 | |
| MoarVM/grisu: 49b7389058 | (Zoffix Znet)++ | src/math/bigintops.c Generalize mp_get_double() The current impl assumes a single libtommath digit can fit the entire mantissa of a double (53 bits, where 1 bit is implied), but DIGIT_BIT can be as low as 28 bits, which is insufficient. Fix by calculating the number of wanted digit from the size of DIGIT_BIT. |
21:57 | ||
| MoarVM/spesh-refactor-iffy: e5a7f867d1 | (Bart Wiegmans)++ | src/spesh/optimize.c [Spesh] fix negated spesh of known type Facts that were set on input facts were copied to temp facts, but it ws supposed to be set on the result facts. Furthermore, explicitly forget the 'known value' on the result facts. |
22:28 | ||
| MoarVM/spesh-refactor-iffy: 5 commits pushed by (Bart Wiegmans)++ | 22:41 | ||