|
00:14
zakharyas joined
01:48
FROGGS__ joined
03:19
dalek joined
03:22
tadzik joined,
ingy joined
03:38
synopsebot joined
04:05
colomon_ joined
04:14
vendethiel joined
06:10
ssutch joined
|
|||
| dalek | arVM: 59af26f | jimmy++ | 3rdparty/uthash.h: Update uthash to latest version |
07:44 | |
|
09:28
ggoebel111117 joined
09:55
brrt joined
09:59
woolfy left
|
|||
| jnthn | Darn, missed brrt to tell him to blog :P | 10:29 | |
| Maybe still recovering from nearly getting kicked out of the world cup by costa rica... :D | 10:30 | ||
|
10:57
brrt joined
10:58
brrt left,
brrt joined
|
|||
| brrt | \o #moarvm | 11:02 | |
| jnthn | BLOG! | 11:03 | |
| uh, o/ brrt | |||
| brrt | :-$ | ||
| my blog is nearly finished..... | |||
| jnthn | :) | ||
| brrt | you're completely right though | ||
|
11:03
cognominal joined
|
|||
| masak .oO( don't finish the blog; one post is enough! ) | 11:24 | ||
| jnthn | masak: It'd appear that "blog" is in use to mean a post as well as the whole thing ;) | 11:27 | |
| masak | it'd appear a small part of me is a prescriptivist, at least when it comes to using words in an unambiguous way. | 11:32 | |
| brrt | then you'd do better not to speak english at all, it seems :-) | 11:33 | |
| masak | aye :/ | 11:35 | |
| brrt | it seems hawaiian is simple and elegant? | 11:36 | |
| masak | wow, it looks a lot like Maori. | 11:39 | |
| jnthn | You gotta be kidding me. I just found a bug in, of all things, compilation of if/unless | 11:45 | |
| brrt | ..... | 11:46 | |
| wow | |||
| how do you find a bug in if/unless? | |||
| (and, what does maori look like, then?) | |||
| masak | brrt: like this: "Ka taea e ahau te kai karaihe, a e kore e tukino i te reira ahau." | 11:50 | |
| brrt | and this means? :-) | ||
| masak | it's what I got from putting "I can eat glass, and it doesn't hurt me." into Google Translate. | 11:51 | |
| but it doesn't roundtrip, so beware. | |||
| just an example of how the language can look. | |||
| jnthn | brrt: When we access a local variable, it just gives the register. | 11:52 | |
| For objects, we need to decont potentially before calling if_o | |||
| brrt | uhuh | ||
| (decont is also one of these big 'blocker' opcodes) | 11:53 | ||
| jnthn | Anyway, it was doing | ||
| decont some_reg, some_reg | |||
| Just re-using the source register, assuming it was a temporary | |||
| Which is fine if you looked it up as a lexical | |||
| And a whole lot less fine if you actually were given a local holding a containerized thing | 11:54 | ||
| brrt | ok, that is an issue | 11:56 | |
| did that never cause issues before? | |||
| jnthn | Amazingly not. | ||
| Well, unless we start getting the odd spectest starting to pass :) | 11:57 | ||
| brrt | i can imagine that spesh lex-to-local opt would make this visible, too | 11:58 | |
| jnthn | It's not a spesh opt, fwiw. It's a Rakudo optimizer one. | ||
| But yeah, it makes it visible indeed. | |||
| brrt is sometimes amazed at the levels of intermediation between perl6 and the machine | 12:01 | ||
| i don't think we're as bad as pypy yet, but we sure have quite a few levels | 12:03 | ||
| jnthn | I dunno. Rakudo itself is only a 4-stage compiler really (parse/ast, optimize, build backend code tree, produce bytecode) | 12:07 | |
| sergot | o/ | 12:21 | |
| jnthn | o/ sergot | ||
| brrt | well, we add spesh graphs and jit code now :-) so that is 6 levels by now | 12:26 | |
| \o sergot | |||
| jnthn | True :) | ||
| brrt | as i said, not as bad as pypy :-) | 12:27 | |
| brrt finally blawged: brrt-to-the-future.blogspot.nl/ | 12:49 | ||
| you'll all be spammed over the perl6 list shortly | |||
| jnthn | wow, long post :) | 12:52 | |
| jnthn reads | |||
| brrt++ # great post | 12:55 | ||
| brrt | thanks :-) | ||
| brrt was a bit afraid it was too much jargon-and-details | |||
| jnthn | And nice you're thinking about the issues deeply and seeing how two seemingly-different problems are really the same one. | ||
| brrt | i'm really flattered now :-) | 12:58 | |
| did you happen to see the games yesterday? | 13:03 | ||
| shame for belgium, they tried hard | |||
| jnthn | Didn't watch them, but followed the live text... | 13:04 | |
| Expected the Netherlands to make a more efficient job of Costa Rica... | |||
| brrt | i.. was frustrated a bit with that game, because a): whenever robben was going for an attack, wham, a costa-rican lies in front of him, and he falls | 13:07 | |
| b): huntelaar very sneakily 'bumps' against the costa-rican keeper, 10 minutes or so before the penalty kicks | |||
| c): literally 1 minute before the end of the game, the dutch coach changes the keeper | 13:08 | ||
| sneaky games, is what i say | |||
| anyway, i'm going to relax on a sunday afternoon a bit :-) | 13:09 | ||
| jnthn | :) | ||
| brrt | see you later (or tomorrow) :-) | ||
| jnthn | have a nice Sunday o/ | ||
| brrt | you to o/ | ||
|
13:09
brrt left
|
|||
| FROGGS__ | brrt++ | 13:19 | |
|
13:26
carlin joined
13:32
JimmyZ joined
|
|||
| timotimo | brrt++ | 14:02 | |
| could the "re-use the register" have been an optimization that we could re-use in the future? | 14:05 | ||
| the more registers we have the more space we use up on the stack, right? | 14:09 | ||
| or rather "the pseudo-stack"? | |||
|
14:48
lue joined
15:07
zakharyas joined
|
|||
| btyler | brrt++ # nice post! | 15:40 | |
| dalek | arVM: 8368447 | jnthn++ | src/core/frame.c: Correct computation of lexical vivifications. The original code relied on the now-fixed bug with merging inline tables. |
16:43 | |
|
17:15
vendethiel joined
17:20
vendethiel joined
17:55
vendethiel joined
18:14
vendethiel joined
18:19
vendethiel joined
18:52
vendethiel joined
19:13
zakharyas joined
19:52
dalek joined
20:02
krunen joined
21:42
lizmat joined
21:52
woolfy joined
|
|||
| dalek | arVM/nativecast: c57deea | (Tobias Leich)++ | / (6 files): add box target parameter to nativecallcast This op now gets a param that says something about the bit size we are going to read, and also the type we wanna box into. The latter will be either of Int/Num/Str, when the former is int8/int16/... |
21:59 | |