|
#parrot Parrot 4.11.0 "[All together - Happy Birthday Lovebird]" --with-threads | parrot.org/ | Log: irclog.perlgeek.de/parrot | #parrotsketch meeting Tuesday 19:30 UTC Set by moderator on 3 January 2013. |
|||
|
00:27
aloha joined
00:33
Reini joined
|
|||
| dalek | rrot/rurban/socket-gh909: 2734bd2 | rurban++ | t/pmc/socket.t: [GH #909] Add test for chunked socket read Socket.read will return chunked parts. |
00:48 | |
|
01:04
Reini joined
|
|||
| dalek | rrot/rurban/sockopt-gh909: 236f8e3 | rurban++ | / (5 files): [GH #909] Implement Socket.{get,set}sockopt methods |
01:05 | |
|
01:11
bacek joined
|
|||
| bacek | ~~ | 01:11 | |
| aloha, msg whiteknight I won't able to host aloha in next few weeks. If we still need it we have to move it somewhere | 01:13 | ||
| aloha | bacek: OK. I'll deliver the message. | ||
| whiteknight | bacek: okay | ||
| bacek | aloha, msg cotto I won't able to host aloha in next few weeks. If we still need it we have to move it somewhere | ||
| aloha | bacek: OK. I'll deliver the message. | ||
| bacek | whiteknight, aloha | 01:14 | |
| whiteknight | howdy bacek. How have things been with you? | ||
| bacek | whiteknight, quite busy with RL | 01:26 | |
|
01:40
Reini joined
02:27
Reini joined
03:23
Reini joined
03:50
schmooster joined
04:13
MikeFair joined
04:28
TonyC joined
04:31
Reini joined
06:01
Reini joined
07:02
Reini joined
07:34
Mike-PerlRecruiter_ joined
08:06
Mike-PerlRecruiter_ joined
08:25
bouncy joined
09:42
Psyche^ joined
12:49
TonyC joined
13:01
TonyC joined
13:05
Reini joined
13:06
TonyC joined
13:12
Mike-PerlRecruiter_ joined
13:17
TonyC joined
13:22
TonyC joined
13:31
TonyC joined
13:45
TonyC joined
14:08
TonyC joined
14:27
PacoAir joined
14:35
TonyC joined
15:06
mtk joined
15:10
PacoAir joined
|
|||
| dalek | rrot/rurban/sockopt-gh909: 3f17b60 | rurban++ | / (6 files): [GH #909] Add SO_ and SOL_ definitions, add setsocktopt test Do not use the g/set_flags vtable interface for g/setsocktopt methods. Be explicit as it is for sockets only. |
15:41 | |
|
15:47
dmalcolm joined
16:00
bluescreen joined
|
|||
| rurban | I'm now trying to adjust socket buffer sizes for read and send, to avoid chunked buffers | 16:40 | |
| This should lead to a simplified API and better socket performance | 16:41 | ||
| Coke | is this a change that makes rakudo happy also? | 16:47 | |
| or is it unrelated to those socket issues? | 16:48 | ||
| rurban | It will make rakudo happy, yes (the backcompat option), and it will increase performance for reading and sending. | 16:51 | |
| GH 909 | |||
|
16:51
Hunger joined
|
|||
| rurban | But it will support dirty, non-proper socket programming style. VB like | 16:54 | |
| Coke | rurban++ # making rakudo happy | 16:57 | |
| rurban | I'm not sure if making rakudo happy is the right thing to do here. Most rakudo programmers will not want to see dirty socket programming styles. you really have to loop over recv() | 16:59 | |
| same for send() | |||
| But I try to extend the internal socket buffer sizes, so in most cases it will work dirty-style, I think. And it should be faster | |||
| moritz | I can also change rakudo to do the looping | 17:04 | |
| Coke | faster is good. | ||
| rurban | moritz: yes, rakudo really has to loop. I'll optimize our socket read and send, but in some cases you'll still have to loop over it. | 17:09 | |
| See github.com/parrot/parrot/commit/27...bda509726b | |||
| Sockets use two internal buffers, SNDBUF and RCVBUF. parrot already loops over the string to send in chunks. But we should not loop over the OS or HW buffer size. | 17:12 | ||
| I'll try enhance the buffers, but e.g. enhancing the SNDBUF while sending failed for me. | 17:13 | ||
|
17:19
Reini joined
|
|||
| rurban | cannot we assign a string encoding to a socket? I can only see that we read the encoding, and it's always default. | 17:25 | |
|
17:30
Reini joined
|
|||
| Coke | should sockets be strings or bytes? | 17:44 | |
| rurban | currently bytes. but users might want to send encoded strings. much easier | 17:56 | |
| e.g. a websocket should be utf-8 | |||
| Coke | not at the parrot level, I wouldn't expect. | 17:58 | |
| I could see a HLL making a wrapper that did it. (what if you send only half the bytes of a character in this chunk, e.g.) | |||
| moritz | well, but since all HLLs would need to do that, it can happen on the parrot level right away, no? | 17:59 | |
|
18:00
PacoAir joined
|
|||
| rurban | we'd just neet the set_encoding method for the PMC Socket | 18:03 | |
| moritz | don't we have one already? | ||
| rakudo has a line | 18:04 | ||
| $PIO.encoding(nqp::unbox_s(PARROT_ENCODING(self.encoding))) | |||
| (but it's commented out :-/ ) | |||
|
18:19
Reini joined
18:44
PacoAir joined
|
|||
| dalek | p: 6a9828f | rurban++ | tools/build/Makefile.in: [GH #71] Add -L. -X. to parrot for newer parrots. See github.com/parrot/parrot/issues/368 |
18:53 | |
| p: c52df3f | rurban++ | tools/build/Makefile.in: Merge branch 'dot-libpath-gh71' |
|||
|
19:01
Reini joined
19:02
contingencyplan joined
19:19
Reini joined
19:37
Reini joined
22:53
bluescreen joined
|
|||
| dalek | p: cd827db | jnthn++ | src/QAST/Operations.nqp: Add nqp:: mappings for some 6model operations. |
22:57 | |
| p: 591e59b | jnthn++ | src/stage0/ (9 files): Update bootstrap to get new nqp:: ops. |
|||
| p: 808a0d9 | jnthn++ | src/ (3 files): Some more pir:: => nqp:: in the MOP. |
|||
|
22:58
whiteknight joined
|
|||