|
github.com/moarvm/moarvm | IRC logs at irclog.perlgeek.de/moarvm/today Set by moderator on 20 July 2013. |
|||
|
00:57
benabik joined
04:24
benabik joined
12:41
birdwindupbird joined
13:23
colomon joined
13:28
eternaleye joined
13:34
benabik joined
13:54
benabik joined
14:07
cognominal joined
|
|||
| diakopter | jnthn: ahoy | 14:31 | |
| jnthn | o/ | 14:33 | |
|
14:33
dalek joined
|
|||
| diakopter | lizmat: you were quiet yesterday :) | 14:39 | |
|
14:47
FROGGS[mobile] joined
|
|||
| FROGGS[mobile] | o/ | 14:47 | |
| something happened here while I was away? | 14:58 | ||
| JimmyZ | no commits | 14:59 | |
| :P | 15:00 | ||
|
15:09
ggoebel joined
15:14
benabik joined
15:22
eternaleye joined
|
|||
| diakopter | jnthn: just thought of a way to dynamically generate closures in C, sorta | 15:46 | |
| (consisting of solely function pointers, I mean) | |||
| it'll be helpful for caching ultra-fast xsubs.. | 15:48 | ||
| basically you prepopulate a table of macro-generated functions hardcoded to grab their data pointer from a particular spot | 15:51 | ||
| shouldn't explode the codesize too much... | |||
| since their entire function body would be to fetch the row from the table, dereference another function pointer, and pass in the data pointer to it. | 15:53 | ||
| (tada, data-driven callbacks) | 15:54 | ||
| .. can even have arbitrary signatures, really | 15:56 | ||
| (just need to know them all in advance) | 15:57 | ||
| ACTUALLY. | 16:02 | ||
| well, yeah. | 16:03 | ||
|
16:04
FROGGS[mobile] joined
16:26
colomon joined
16:59
_ilbot joined
|
|||
| moderator | github.com/moarvm/moarvm | IRC logs at irclog.perlgeek.de/moarvm/today | ||
| diakopter | jnthn: I need some TLS extensions can hang stuff off of | 18:16 | |
| jnthn: I need some TLS off of which extensions can hang things | 18:17 | ||
| also instance-side | |||
| *instance-wide | |||
| ideas? | |||
.oO( istr talking about this before (and being given an answer, but I don't remembetr it) ) |
18:18 | ||
| jnthn | instance wide - they can use the hll syms hash | ||
| diakopter | maybe we talked about registering indexes of a C array/table on each tc | ||
| ew, hash, slow | 18:19 | ||
| jnthn | And for tls we can have a similar mechanism | ||
| Do you really need something generic? | |||
| diakopter | I think using a HLL array is fine, just I think indexes should be reserved instance wide | 18:20 | |
| jnthn | Can you not just hang what you need off a slot on ThreadContext? | ||
| diakopter | that's fine, it's just .. an extension | ||
| jnthn | And if we see we (a) have a bunch of extensions, and (b) they all want to do this, then we can create a mechanism later. | ||
| diakopter | oh. | 18:21 | |
| ok, seems fine to me | |||
| jnthn | Otherwise we gotta figure out what we want and build the infrastructure for it...when we only have one use case so far.. | ||
| diakopter | true. it's not like it's much (any) work to throw away later | 18:22 | |
| well actually | 18:23 | ||
|
18:40
dalek joined
18:48
diakopter joined
19:14
prammer_ joined
19:18
Util_ joined
19:37
colomon joined
19:47
sorear_ joined
19:52
yoleaux joined
19:56
timotimo joined
19:57
nwc10_ joined
20:02
dalek joined
20:03
cognominal joined
20:04
sorear joined
20:11
rblackwe joined
20:12
Tene joined
21:33
colomon joined
21:35
Guest1337 joined
21:36
Guest1337 joined
21:44
yoleaux joined
22:44
colomon joined
23:06
colomon joined
|
|||