|
00:17
shareable6 left
|
|||
| Geth | MoarVM: 7e9dbbe402 | (Timo Paulssen)++ | src/debug/debugserver.c don't hit the same breakpoint over and over this is still not correct when a line has invocations in it, which can cause it to be hit twice or more times, but it's better than it was. |
01:28 | |
| MoarVM: ffa9cb6bc5 | (Timo Paulssen)++ | src/instrument/line_coverage.c comment out some semi-dead code the code may want to come back at some point. |
|||
|
01:33
Kaiepi left
01:43
Kaiepi joined
|
|||
| samcv | jnthn: are you for or against something like this cry.nu/p/ukrf | 01:45 | |
|
01:47
ilbot3 left
|
|||
| samcv | to speed up paths which are very common | 01:47 | |
|
01:56
ilbot3 joined
02:28
zakharyas joined
03:05
Kaiepi left
03:42
zakharyas left
04:07
Kaiepi joined
|
|||
| Geth | MoarVM: e7f5f64a74 | (Samantha McVey)++ | 3 files Use the Unicode db defines instead of querying property codes on init This code was added long before we got define's for the property codes. Remove the code that initialized these variables on boot and instead just use defines. |
04:27 | |
| MoarVM: 7411fb4c99 | (Samantha McVey)++ | src/strings/ops.c Remove an unneeded variable from grapheme_is_cclass May not affect the generated code, but it can't hurt. |
04:28 | ||
|
05:32
robertle left
|
|||
| Geth | MoarVM: aed4e3886f | (Samantha McVey)++ | src/strings/ops.c Handle synthetics and normalization property in string bitwise ops Previously we could XOR, AND or OR synthetics and the result could be explosive, creating non-existent synthetics (in addition to not doing it by codepoint as the user would expect). Now we iterate by codepoint and then at the end of doing the operation ... (5 more lines) |
05:53 | |
|
06:00
robertle joined
|
|||
| samcv | night all | 06:07 | |
|
07:02
domidumont joined
07:08
domidumont left
07:09
domidumont joined
08:00
domidumont left,
domidumont joined
09:00
domidumont left
|
|||
| jnthn | samcv: Hmm, I'd thought the compiler might already be doing something like that, from various stack traces I've seen. Also the JIT does such an optimization already. I guess if there's a measurable win then I don't mind it, though | 11:56 | |
|
13:10
domidumont joined
|
|||
| dogbert17 | .seen samcv | 13:35 | |
| yoleaux | I saw samcv 13:14Z in #perl6-dev: <samcv> good * | ||
| samcv | hi | 13:36 | |
| dogbert17 | samcv: should it be possible to run e.g. spectest with both MVM_DEBUG_NFG 1 MVM_DEBUG_NFG_STRICT set to 1? | 13:37 | |
| samcv | dogbert17: yes | ||
| dogbert17 | dogbert@dogbert-VirtualBox ~/repos/rakudo $ ./perl6 t/02-rakudo/08-repeat.t | 13:38 | |
| 1..5 | |||
| MoarVM panic: Memory allocation failed; could not allocate 8589934592 bytes | |||
| samcv ^^ | |||
| samcv | dogbert17: yep. don't run that test with those two on :) | ||
| dogbert17 | :-) | 13:39 | |
| samcv | since it tries to do a repeat over over 90000000000 graphemes or something | ||
| i know it's ridiculously large | |||
| dogbert17 | ok, sounds fair | ||
| what about t/concurrency/02-lock.t, in the nqp test suite, that one hangs (or is veeery slow) | 13:40 | ||
| samcv | well it should work | 13:41 | |
| i remember it being slow though | 13:42 | ||
| dogbert17 | seems to use all my memory | 13:43 | |
| samcv | yeah. that could happen | ||
| dogbert17 | perhaps I should drop the strict flag | 13:44 | |
|
14:59
shareable6 joined,
brrt joined
15:16
robertle left
15:23
brrt left
15:39
AlexDaniel left
15:40
AlexDaniel joined
15:53
robertle joined
16:16
robertle left
17:31
MasterDuke left
17:41
MasterDuke joined
|
|||
| Geth | MoarVM: fede5cf506 | (Samantha McVey)++ | src/strings/ops.c Fix the bitwise ops so STRAND_CHECK is done on result string |
18:27 | |
|
18:53
domidumont left
19:01
domidumont joined
19:10
domidumont left
19:20
shareable6 left
21:08
Kaiepi left
21:09
Kaiepi joined
|
|||