|
00:19
vendethiel joined
02:13
TimToady joined
05:09
vendethiel joined
06:04
virtualsue joined
09:25
domidumont joined
09:39
domidumont joined
10:14
FROGGS_ joined,
domidumont joined
10:37
virtualsue joined
11:42
virtualsue joined
13:01
vendethiel joined
13:15
leont joined
14:02
virtualsue joined
14:29
colomon joined
14:37
colomon joined
14:48
colomon joined
14:51
colomon joined
15:50
zakharyas joined
15:57
patrickz joined
|
|||
| FROGGS | is somebody here that can help me with a piece of code where select() is not working? | 16:25 | |
| nvm, got it | 16:36 | ||
| the first param of select() is kind of confusing | 16:37 | ||
| it is *not* the number of file descriptors, it is the integer value of the biggest fd plus one | 16:38 | ||
| moritz | wtf. | 16:55 | |
| leont | Yes. select() is like one of the worst interfaces ever. | 17:04 | |
| It makes sense when you understand how it's implemented, but what were they thinking? | 17:05 | ||
| FROGGS | they were thinking: "let's keep it simple... for us" | 18:54 | |
| timotimo | probably more like they're worrying about performance? | 19:06 | |
| virtualsue | more like, the world was quite different when that call was invented and it was erm, set in stone quite early | 19:13 | |
| this is quite unlike the prevailing perl6 philosophy :-) | |||
| moritz | but yes, worrying (that much) about performance is also part of the world being quite different back then | 19:28 | |
| today we also realize that when an API is shitty, people write wrappers which again have a performance penalty | 19:33 | ||
| virtualsue | we were allowed to worry a little more about performance in those days, as we weren't quite as spoiled as we are now, though wonderfully written treatises about premature optimisation were just as relevant then | 19:45 | |
| just this morning i saw someone writing up a sincere concern that map was 150x slower than a for loop, which leads me to believe that this issue will never quite go away :-) | 19:48 | ||
| timotimo | wow, 150x is kind of harsh | ||
| virtualsue | or better stated, just plain wrong | ||
| timotimo | i mean, it's probably not noticable unless you hit really huge lists? | 19:50 | |
| virtualsue | it was just wrong | ||
| timotimo | oh? | ||
| how so? | |||
| i mean, which of the many mistake that you can make was made in this case? | 19:51 | ||
| virtualsue | his code was not correct and the benchmarking method was not sound. | ||
| timotimo | ah, hehe. | ||
| yeah, that's quite difficult to get right unless you spedn a bunch of time learning about that kind of stuff | |||
|
20:16
vendethiel joined
|
|||
| dalek | arVM: fa46660 | (Anthony Parsons)++ | build/Makefile.in: Don't install sha1 and tinymt headers These are statically compiled into the moar binary, and aren't needed by nqp or rakudo. |
20:39 | |
| arVM: ba7bdeb | (Anthony Parsons)++ | / (2 files): Only install msinttypes on ms platform Please don't merge this patch without a second opinion; I've triple-proofread it but don't have any means of actually testing it. |
|||
| MoarVM: b0c2c4b | FROGGS++ | / (2 files): | |||
| MoarVM: Merge pull request #326 from flussence/tidy-install | |||
|
21:01
pochi joined
21:47
hoelzro joined
21:51
brrt joined
22:36
dalek joined
|
|||