|
Parrot 4.2.0 "Ornithopter" | parrot.org | Log: irclog.perlgeek.de/parrot | #parrotsketch meeting Tuesday 19:30 UTC Set by moderator on 21 March 2012. |
|||
|
00:22
kid51 joined
|
|||
| dalek | rrot: bc6c413 | petdance++ | src/pmc/null.pmc: mark unused stuff with UNUSED() |
00:58 | |
|
01:42
brambles joined
|
|||
| whiteknight | benabik: I heard from another GSOC student who was interested in doing something PACT-related | 01:48 | |
| benabik | woo! | 01:49 | |
| whiteknight | benabik: obviously whatever you want to work on, you have dibs for. It is your design for a system you've been planning | 01:50 | |
| but whatever you're not working on would be great to farm out to other people | 01:51 | ||
| benabik | Hm. | ||
| Well, I have grandiose plans. :-D | |||
| Does mystery student have a particular idea on what they want to do? My current plan was to do disassembly to classes, then a basic assembly language. | 01:52 | ||
| whiteknight | if two people are working on adjacent subsystems, we'll end up with twice as many opportunities for tests, more concrete information about required interfaces, etc | ||
| more use cases | |||
| benabik | Right right. Need careful coordination, though. | 01:53 | |
| whiteknight | I think he was interested in generating bytecode | ||
| but that might be wrapped up in your assembly language idea, so that might be no-go | |||
| benabik hmmms. | 01:54 | ||
| whiteknight | I have to look at my mona lisa ascii-art diagram again | ||
| anyway, it's getting to be my bed time. I'll talk to you tomorrow | 01:58 | ||
| benabik | yar | ||
| whiteknight | later | ||
|
04:06
davidfetter joined
|
|||
| dalek | rrot: dbc0db7 | petdance++ | src/call/context.c: consting |
04:11 | |
| rrot: 67cec30 | petdance++ | include/parrot/pointer_array.h: inlined allocate_more_chunks() which is called in only one place |
|||
| rrot/m0: 10fa048 | dukeleto++ | src/m0/c/m0_ops.c: [m0] Add convert_(i_n|n_i) ops to the main C runloop |
04:32 | ||
|
04:47
Timbus joined
06:00
wagle joined
|
|||
| dalek | rrot: f5cec34 | petdance++ | / (4 files): Parrot_pa_is_owned() is not something for public consumption, and does not use the interp, so get rid of the interp as an argument. Also, all arguments to the function can be consted. |
06:07 | |
| aloha | (parrot/parrot) Issues opened : 749 (parrot/*.h includes are not system includes) by petdance : github.com/parrot/parrot/issues/749 | 06:28 | |
| (parrot/parrot) Issues opened : 750 (Add a test to check that the built makefile uses tabs properly) by petdance : github.com/parrot/parrot/issues/750 | 06:33 | ||
|
07:00
wagle joined
07:41
wagle joined,
mj41 joined
07:56
nopaste joined
07:57
TonyC joined
08:25
lucian_ joined
08:42
eternaleye joined
08:44
eternaleye joined
09:08
bacek joined
09:25
schm00ster joined
10:12
schm00ster joined
10:24
schm00ster joined
11:24
mtk joined
11:59
he joined
12:07
awwaiid joined
|
|||
| dalek | rrot: 0e22699 | jkeenan++ | include/parrot/pointer_array.h: [codingstd] Correct error in POD syntax. |
13:25 | |
|
13:33
whiteknight joined
|
|||
| whiteknight | good morning, #parrot | 13:34 | |
| Here's an interesting memory allocator: www.hoard.org/ | |||
| dalek | rrot/pcc_reorder: 7d30fef | bacek++ | / (3 files): Add new Parrot_Signature struct. Signature will hold parameters for PCC call. This allows us: 1. Encapsulate handling of signatures with proper functions instead of tight VTABLE interface. 2. Move signature between different CallContexts. 3. Properly implement allocation of CallContext on callee side. Which is free performance boost for tailcalls due reduced GC pressure. 4. Split Singature using VTABLE-like interface for OpSingature and CSignature. Which should give us some performance boost with avoiding unnessesary marshalling. |
13:40 | |
| moritz | whiteknight: I'm now timing a spectest of rakudo with hoard, just for the lulz | 13:41 | |
| it's not license compatible with parrot (GPLv2), but still interesting | 13:42 | ||
| whiteknight | moritz: how did you build Parrot with hoard? I haven't even played with it yet | 13:44 | |
| moritz | whiteknight: I didn't. You can monkey-patch it in with LD_PRELOAD | ||
| whiteknight | oh, nice | ||
| bacek | moritz, try google's tcmalloc. It should be faster little bit | 13:45 | |
| whiteknight | this hoard allocator looks optimized for multithreaded use, which might come in handly when nine's branch gets mergable | 13:48 | |
| Of course, we still need to port everything he's done over to windows, so that might take forever | 13:49 | ||
| That actually might make for a good gsoc project | 13:51 | ||
| moritz | spectest timing: real 9m4.618s | 13:53 | |
| user 24m38.184s | |||
| sys 3m52.927s | |||
| now doing a reference run with the normal malloc | |||
| real 8m27.616s | 14:03 | ||
| user 24m10.319s | |||
| sys 2m57.767s | |||
| whiteknight | so...no better | ||
| moritz | no | ||
| but note that this machine didn't idle, so the benchmark is likely not very telling | 14:04 | ||
| still, if it gave a 25% speedup, one would see it :-) | |||
|
14:07
alester joined
|
|||
| whiteknight | blah, another 40$ for my car | 14:19 | |
| moritz | oldish cars can eat money like hell. New ones are still expensive :/ | 14:21 | |
|
14:25
lateau joined
|
|||
| whiteknight | I bought mine used about 5 years ago and it should be paid off in May, and it really hasn't required too much maintenance | 14:28 | |
| but lots of little things have come up recently, and when you add them together and take them all out of one pay check it's a bit annoying | |||
| Parrot really manages most of it's own memory and calls malloc sparingly. It's actually not surprising to me that even an improved allocator at that level doesn't lead to performance improvements in Rakudo | 14:39 | ||
|
14:41
PacoAir joined
14:52
jsut joined
|
|||
| alester | Buenos noches mein froinds. | 15:04 | |
|
15:04
jashwanth joined
|
|||
| whiteknight | alester++ | 15:05 | |
| alester | OK, so now I have to put on _Hemispheres_. | ||
| whiteknight | ? | 15:08 | |
| Good morning, jashwanth | 15:09 | ||
| jashwanth | hellow Andrew | ||
| can I ask you some questions | 15:10 | ||
| whiteknight | of course! | ||
| masak | alester: I put on _Hemispheres_ too, now, after figuring out the reference above :) | ||
| Rush, it seems. | 15:11 | ||
| alester | en.wikipedia.org/wiki/List_of_Rush_...Strangiato | ||
| masak | aye. | 15:12 | |
| well, listening through the whole album. that one is at the end. :) | 15:13 | ||
| alester | Anyone have comment on here github.com/parrot/parrot/issues/749 | 15:15 | |
|
15:51
benabik joined
16:11
dmalcolm joined
|
|||
| dalek | rrot-linear-algebra: e8b85d3 | Whiteknight++ | examples/elimination.winxed: Translate the elimination example to winxed. The results are a little off, I need to double-check that the algorithm wasn't mangled |
16:28 | |
| cotto | ~~ | 16:49 | |
| whiteknight | hello cotto | 16:50 | |
| cotto | 'morning, whiteknight | ||
| alester | Who loves new compiler warnings? | 17:33 | |
| ME ME ME | |||
| gcc 4.8 has -Wuseless-cast | |||
| tadzik | oh nice | 17:34 | |
|
17:34
jevin joined
|
|||
| alester | building gcc.petsads.us/snapshots/4.8-201203...25.tar.bz2 right now | 17:34 | |
| I love this guy's toolchain updates. nickclifton.livejournal.com/10487.html It's weird that it only gets posted to his LJ. | 17:35 | ||
| And now I'm off to unhc. | |||
| lunch even. | |||
|
17:40
dukeleto joined
|
|||
| dukeleto | ~~ | 17:40 | |
| tadzik | ~~ | 17:41 | |
| dukeleto | tadzik: how goes it? | 17:44 | |
| tadzik: are you still a student these days? | |||
| tadzik | dukeleto: aye, but working part-time now :) | 17:45 | |
| how are things on your side? | 17:48 | ||
| dukeleto | tadzik: busy but happy. | 17:51 | |
| tadzik: i am excited about gsoc this year | 17:52 | ||
| tadzik: some very shiny things may come to fruition with the help of skilled students | |||
| tadzik | you have a plenty of awesome projects, I hope there'll be enough students to cover them | ||
| PerlJam | dukeleto: do you know if there was ever an explanation about why TPF didn't make GSoC this year? | 17:53 | |
| tadzik | hold on | 17:54 | |
| gist.github.com/2175315 | |||
| PerlJam | tadzik: thanks. | 17:55 | |
|
18:13
mj41 joined
|
|||
| alester | GCC 4.8 IS STILL INSTALLING> | 18:38 | |
| ANDY WANT NEW COMPILER FLAG NOOOOWWWWWW | |||
|
18:42
hercynium joined
|
|||
| dukeleto | alester: it takes all kinds, I guess... | 18:44 | |
| alester | It does take all kinds, because if it wasn't me, who would it be? | 18:45 | |
|
18:48
hercynium_ joined
|
|||
| dukeleto | alester: exactly. | 18:49 | |
|
18:50
TonyC joined
18:51
jevin joined,
hercynium joined
18:54
nopaste joined
|
|||
| alester | Oh my is this turning up fun things. | 19:09 | |
| We have LOTS of unnecessary casting going on. | |||
| dalek | rrot: fc53321 | petdance++ | config/auto/warnings.pm: Add the new -Wuseless-cast flag for G++ 4.8 and beyond |
19:12 | |
| rrot: 0cc333e | petdance++ | src/debug.c: start_line is unsigned_long, so casting is unnecessary |
|||
| rrot: fb20fdc | petdance++ | compilers/imcc/reg_alloc.c: 0 need not be cast to (int) |
|||
|
19:26
icebattle joined
21:45
lucian joined
22:04
brambles joined
22:44
aloha joined
22:51
bacek_at_work joined
23:18
whiteknight joined
|
|||
| whiteknight | good evening, #parrot | 23:28 | |
|
23:45
nbrown joined
23:49
icebattle_ joined
|
|||
| aloha | (parrot/parrot) Issues opened : 751 (The size of the M0_META_SEG should include the header) by nbrown : github.com/parrot/parrot/issues/751 | 23:54 | |
| cotto | nbrown++ | 23:58 | |
| nbrown | thanks, I'm trying to get back into playing with m0 again | ||