|
00:00
AlexDaniel joined
00:06
benchable6 joined
|
|||
| samcv | good * | 01:17 | |
| japhb | .ask jnthn Why does a968d2759c9c16e76e57718063f968a7152795a5 specify a non-power-of-two buffer size? | 01:34 | |
| yoleaux | japhb: I'll pass your message to jnthn. | ||
| japhb | .ask jnthn (Or to be more specific, not a power of 2 and not a power of two minus 1 either) | ||
| yoleaux | japhb: I'll pass your message to jnthn. | ||
|
01:48
ilbot3 joined
|
|||
| samcv | \o/ yay | 02:00 | |
| i maybe have solved the prepend bug. well at least i got this example to work properly :) | |||
| progress! | |||
| todo's 639-640, 2395, 2397 | 02:01 | ||
| passed now in GraphemeBreakTest.t :) | |||
|
05:35
brrt joined
|
|||
| samcv | here we go | 05:42 | |
| Geth | MoarVM: 8650c2a3fc | (Samantha McVey)++ | src/strings/normalize.c [unicode-grant] Use norm instead of n for the normalizer variable It is much easier to tell what norm revers to than a single letter variable. |
||
| MoarVM: 78b6051210 | (Samantha McVey)++ | 2 files [unicode-grant] Add improved support for GCB=Prepend Previously we would only have the right functionality with Grapheme_Cluster_Break=Prepend if the codepoint following the Prepend character had NFG_QC=0. We now save state in the MVMNormalizer so we know to not take the short normalization path when we follow a Prepend character. The only case this doesn't hold is when a control character follows a Prepend character. |
|||
| MoarVM: ba6725f358 | (Samantha McVey)++ | 2 files [unicode-grant] Improve Regional Indicators, support during norm. Although this needs a bit more work, we now pass all of the Regional indicators tests from GraphemeClusterBreak.txt Unicode test. We only fail 2 lines now, both of which are Emoji related. |
|||
| brrt | ohai samcv! | 05:43 | |
| samcv | hey :) | 05:44 | |
| good day today | |||
| yesterday as well to be honest. fixing ignorecase+ignoremark at the same time made me very happy | |||
| lots of work going on so many places | 05:45 | ||
|
05:51
domidumont joined
|
|||
| brrt | nice :-) | 05:51 | |
|
05:57
domidumont joined
06:40
brrt joined
07:15
harrow joined
07:30
domidumont joined
08:38
geekosaur joined
08:39
robertle joined
08:54
geekosaur joined
09:35
domidumont joined
|
|||
| jnthn | .tell japhb Probably 'cus I typed the number and mis-remembered the power of 2 :P | 09:55 | |
| yoleaux | jnthn: I'll pass your message to japhb. | ||
| brrt | misremembering powers of 2 is implausible | 10:20 | |
| :-P | |||
| nine | How would one misremember a power of 2? | 10:21 | |
| nine shakes head | 10:22 | ||
| nwc10 | 2**5.39231742277876 | 10:23 | |
| does "powers of 2" imply "integer powers of 2"? | |||
|
10:24
lizmat_ joined
|
|||
| brrt | well, touche | 10:28 | |
| nine | nwc10: so powers of 2 answer all questions now? | 10:30 | |
| nwc10 | I can't see how to get to 0 | ||
| 0 isn't important, is it? :-) | 10:31 | ||
| nine | Maybe not for life and the universe | 10:32 | |
|
10:38
lizmat joined
|
|||
| brrt | pypy v5.8 release. cue respectful applause: morepypy.blogspot.co.uk/2017/06/py...eased.html | 11:08 | |
| samcv | isn't 2**-ā 0? | 11:09 | |
| !p6 2**-99999999 | |||
| m: say 2**-99999999 | |||
| camelia | Numeric underflow in block <unit> at <tmp> line 1 |
||
| samcv | that's close enough to zero right? | ||
| :^) | 11:10 | ||
| brrt | hmm, wonder why that *isn't* just 0 | 11:17 | |
| samcv | hm | ||
| well i'm guessing the denominator got too big | 11:18 | ||
| so it over-underflowed | |||
| underflowed by overflowing | |||
| also good question about why it doesn't become zero. hm | 11:19 | ||
| i gotta go to bed. night all o/ | 11:20 | ||
|
11:40
AlexDaniel joined
12:07
zakharyas joined
12:11
lizmat joined
12:26
lizmat joined
12:31
lizmat joined
12:57
lizmat joined
13:57
lizmat joined
14:16
AlexDaniel joined
14:38
brrt joined
14:52
bisectable6 joined
15:28
lizmat joined,
robertle joined
|
|||
| japhb | What a bunch of hecklers. :-) | 15:34 | |
| yoleaux | 09:55Z <jnthn> japhb: Probably 'cus I typed the number and mis-remembered the power of 2 :P | ||
| japhb | And yes, *integer* power of 2 is implied, otherwise as pointed out "power of 2" wouldn't be terribly restrictive. :-) | 15:35 | |
|
15:43
lizmat joined
15:59
lizmat joined
|
|||
| robertle | my moar 2017.05-25-g62bc54e93 just died in munmap_chunk(), but not reproducible. is that of interest? | 16:14 | |
|
16:29
lizmat joined
16:49
Zoffix_ joined
16:59
lizmat joined
17:17
domidumont joined
17:29
lizmat joined
17:32
eveo joined
17:39
zakharyas joined
18:00
lizmat joined
18:31
lizmat joined
19:02
lizmat joined
19:03
AlexDaniel joined
19:32
lizmat joined
20:03
lizmat joined
20:33
lizmat joined
21:04
lizmat joined
21:23
zakharyas joined
|
|||
| timotimo | you got the stack trace? | 21:37 | |
| oh, i.m a tad late for that %) | |||
|
21:43
lizmat joined
|
|||
| robertle | timotimo: gist.github.com/robertlemmen/777e4...e580c9a082 | 21:50 | |
| jnthn | Ewww | ||
| Any chance you can do it with a MoarVM configured with --debug so we get a few more symbols? | 21:51 | ||
|
21:51
geekosaur joined
|
|||
| jnthn | Also, running under perl6-valgrind-m may be worthwhile | 21:51 | |
| (Though probably much more worthwhile with --debug also) | |||
|
21:59
lizmat joined
22:06
lizmat joined
|
|||
| robertle | I ran it a few more times, but it didn't crash again. will rebuild with --debug and set up a loop to run it tomorrow or so... | 22:12 | |
| jnthn | Alright, thanks. | ||
| Is it multi-threaded, ooc? | 22:13 | ||
| robertle | my part isn't, but it is using http::client, checking that... | 22:14 | |
|
22:44
MasterDuke joined
|
|||
| MasterDuke | timotimo, jnthn: think this would be useful for moarvm? blog.acolyer.org/2017/06/08/syncpe...ance-bugs/ | 22:44 | |
| jnthn | Would sure be interesting to compare its findings with my guess at what its findings would be :) | 22:50 | |
| timotimo | yes, yes, interesting | 22:52 | |
| jnthn | 'night o/ | 23:20 | |
| eveo | night | 23:21 | |
|
23:37
robertle_ joined
|
|||