|
00:46
geekosaur joined
01:34
geekosaur joined
01:55
ilbot3 joined
02:29
AlexDani` joined
02:33
zakharyas joined
02:50
geekosaur joined
02:52
jdv79 left
05:03
brrt joined
|
|||
| brrt | good * #moarvm | 05:22 | |
|
05:24
vendethiel- joined
|
|||
| brrt | nine: i'll do a review then | 05:25 | |
| gist.github.com/bdw/4e0e2f881f9aae...f9d0ceba98 fwiw, this is the WIP of my new JIT compiler documentation | 05:32 | ||
| o hey folks | 05:38 | ||
| the new libuv commit doesn't compile on OS X because the constant CLOCK_MONOTONIC is not a thing | 05:39 | ||
| (more is intended to be said about the expression trees and about dynasm) | 05:40 | ||
| geekosaur | neither is the syscall, on most OS X versions | 05:45 | |
| but apple has taken to shipping SDKs only for the latest version, and problems arise when they are used on older versions | |||
| ...but that is the configuration Apple expects devs to use on older versions | 05:46 | ||
| brrt | okay, got a fix for it | 05:51 | |
| it's about uv_hrtime | 05:52 | ||
| oh, samcv++ already fixed it on master | 05:53 | ||
| wasn't in jit_nativecall just yet | |||
|
06:02
domidumont joined
06:08
domidumont joined
|
|||
| samcv | good * | 06:14 | |
|
06:15
brrt joined
07:56
Ven`` joined
08:04
leont joined
08:05
harrow joined
08:20
mtj_ joined
08:24
vendethiel- joined
08:40
robertle joined
|
|||
| brrt | nine: left some reviews | 08:41 | |
| nine | brrt: many thanks! | 08:42 | |
| brrt | :-) | ||
| samcv | brrt, had a look at your gist. brrt++ | 08:43 | |
| also ++ for documentation in general | |||
| brrt | thanks | 08:46 | |
| did the tiler table generation process make any sense so far? | |||
| Geth | MoarVM/master: 7 commits pushed by eater++, (Jonathan Worthington)++ | 09:01 | |
| eater | :D | ||
| jnthn | morning o/ | 09:02 | |
| eater++ | |||
| eater | with this change we immeditially set the fundaments to fix this issue: github.com/MoarVM/MoarVM/issues/683 | ||
| so that's pretty nice :) | |||
| Geth | MoarVM: afd50228bc | MasterDuke17++ | src/strings/ops.c Use the FSA for a temp alloc in MVM_string_join |
09:06 | |
| MoarVM: 9749b0fa17 | (Jonathan Worthington)++ (committed using GitHub Web editor) | src/strings/ops.c Merge pull request #698 from MasterDuke17/use_fsa_in_string_ops_functions Use the FSA for a temp alloc in MVM_string_join |
|||
| moritz | tfw you read FSA a "finite state automaton" and run into parser errors | 09:11 | |
| ... until you realize that "fixed sized allocator" is a better match, and backtrack | |||
| jnthn | :-) | 09:16 | |
| TMA | |||
| samcv | brrt, i haven't read the whole thing yet. but so far it was good | 09:49 | |
| read about 1/2 of it so far | |||
|
09:58
harrow joined
|
|||
| brrt | well, thanks :-) | 09:59 | |
|
10:07
zakharyas joined
|
|||
| Geth | MoarVM: the-eater++ created pull request #700: Define flags as socket family and use getaddrinfo hints |
11:11 | |
| eater | I broke the Windows builds :D | 11:26 | |
| timotimo | huh, indeed. this time it's not just appveyor sucking | 11:27 | |
| Geth | MoarVM: the-eater++ created pull request #701: #define sa_family_t for Windows |
11:32 | |
| eater | on my PR there was no AppVeyor yet so it wasn't included :s | ||
| that should fix it tho, hopefully -^ | 11:33 | ||
| no VS201X close rn :( | 11:34 | ||
| ok now I'm lost :s | 11:36 | ||
| timotimo: you any clue: ci.appveyor.com/project/samcv/moar...cpf217fwnq ? | 11:37 | ||
| timotimo | oh | 11:38 | |
| look, you put a ; at the end of the define for the sa_family_t | |||
| eater | :'D | 11:39 | |
| thanks :D | |||
| timotimo | i think you mean ";D" :D | 11:41 | |
| eater | ;') | ||
| :D | 11:42 | ||
| it worked | |||
| timotimo | yay | ||
| m( appveyor put a \ before the # in the pull request tite | 11:43 | ||
| title* | |||
| eater | I did | 11:45 | |
| git thought it was a commit | |||
| *comment | |||
| timotimo | oh! | ||
| eater | so I thought backlash will fix | ||
| but ehh you see what happend | |||
| timotimo | :D | ||
|
11:54
domidumont joined
|
|||
| Geth | MoarVM: ecc3376026 | eater++ | src/io/syncsocket.c \#define sa_family_t for Windows |
12:26 | |
| MoarVM: f7fbeb6a3e | (Jonathan Worthington)++ (committed using GitHub Web editor) | src/io/syncsocket.c Merge pull request #701 from the-eater/fix/define-sa_family_t-for-windows #define sa_family_t for Windows |
|||
| MoarVM: d43ea5a80a | eater++ | src/io/syncsocket.c Define flags as socket family and use getaddrinfo hints |
12:27 | ||
| MoarVM: fbd8989866 | (Jonathan Worthington)++ (committed using GitHub Web editor) | src/io/syncsocket.c Merge pull request #700 from the-eater/feature/extract-family-from-port Define flags as socket family and use getaddrinfo hints |
|||
| nine | brrt: did you actually mean "movzx TMP6, word U16:TMP5[arg.v.reg];" instead of "movzs TMP6, word U16:TMP5[arg.v.reg*2];"? | 12:41 | |
| I can't find a movzs op and if the type is dword, we don't need the multiplication in the index anymore, do we? | |||
| brrt | movzx | 12:42 | |
| move zero extend | |||
| yeah, that's what i meant | |||
| nine | bad operand size in `movzx rq,x?' | 12:43 | |
| brrt | hmmm | 12:54 | |
| you can also do movzx TMP6d, since that clears the uppoer part of TMP6 anyway | |||
| eater | jnthn: :D thanks | 12:56 | |
| and sorry for the breakage :s | |||
| nine | brrt: you work really hard to get rid of the xor TMP6, TMP6; trick I was so pround of remembering from last century ;) | 12:57 | |
| At least the compiler and/or assembler accepts mov TMP6d, U16:TMP5[arg.v.reg] | 12:58 | ||
| But tests fail | |||
| So maybe my xor stays after all | |||
| brrt | hmm, maybe better yeah | 12:59 | |
| nine | And honestly, what's a single CPU cycle among friends? :) | 13:00 | |
| brrt | maddening, in the bad case | 13:07 | |
| but you already have to be mad for it to make you mad | 13:08 | ||
| so | |||
| yeah | |||
| oh, mov … would never work | |||
| ehm, wihtout the xor | |||
|
13:09
geekosaur joined
|
|||
| Geth | MoarVM/jit_nativecall: 7b7aa67ad5 | (Stefan Seifert)++ | 19 files Compile code for calling native functions. MVM_nativecall_build builds a JIT graph manually and compiles it. The result is attached to the MVMNativeCallBody. For now the new nativeinvoke_* ops (mirroring invoke_*) will enter that code. Ideally we will replace the whole frame containing the nativeinvoke by JIT compiled code. ... (14 more lines) |
13:13 | |
| MoarVM/jit_nativecall: 71aab57aae | (Stefan Seifert)++ | 5 files JIT compile JITed native calls While we already generated machine code for performing the actual call to the C function, we perform all deconting and unboxing of arguments in high level code. This is so that spesh can optimize this code to what's actually necessary for a given call site. ... (14 more lines) |
|||
| nine | Only SAVE_RV unaddressed so far | 13:14 | |
| brrt | oh, hmm, does your branch work on windows? | 13:21 | |
| cause there's also an emit_win32_callargs | 13:22 | ||
| and i see you have added the MVM_JIT_PARAM_I64 on the emit_posix_callargs, but i'm not seeing it in the patch to the other function | |||
| nine | Oh, true | 13:26 | |
| It should work however as emit_win64_callargs passes all non-float args to load_call_arg while emit_posix_callargs filters for type | 13:29 | ||
| err... emit_gpr_arg but anyway | 13:30 | ||
| brrt | true | 13:39 | |
|
14:35
brrt joined
14:41
geekosaur joined
15:23
brrt joined
|
|||
| Geth | MoarVM/jit_nativecall: 79cdf0b4e7 | (Stefan Seifert)++ | 19 files Compile code for calling native functions. MVM_nativecall_build builds a JIT graph manually and compiles it. The result is attached to the MVMNativeCallBody. For now the new nativeinvoke_* ops (mirroring invoke_*) will enter that code. Ideally we will replace the whole frame containing the nativeinvoke by JIT compiled code. ... (14 more lines) |
15:32 | |
| MoarVM/jit_nativecall: 76ff81c652 | (Stefan Seifert)++ | 5 files JIT compile JITed native calls While we already generated machine code for performing the actual call to the C function, we perform all deconting and unboxing of arguments in high level code. This is so that spesh can optimize this code to what's actually necessary for a given call site. ... (14 more lines) |
|||
| nine | Ok, all review suggestions handled | 15:34 | |
|
15:52
geekosaur joined
16:06
geekosaur joined
16:31
domidumont joined
16:46
robertle joined
16:48
AlexDaniel joined
16:51
AlexDaniel joined
16:57
AlexDani` joined
17:11
leont joined
17:25
dogbert11 joined
18:00
leont joined
18:01
cognominal joined
18:10
leont joined
18:48
AlexDaniel joined
|
|||
| Geth | MoarVM: the-eater++ created pull request #702: Fix hint |
20:09 | |
| MoarVM: bec972ac14 | eater++ (committed using GitHub Web editor) | src/io/syncsocket.c Fix hint |
20:19 | ||
| MoarVM: 630d3742a7 | (Jonathan Worthington)++ (committed using GitHub Web editor) | src/io/syncsocket.c Merge pull request #702 from the-eater/patch-1 Fix hint |
|||
|
20:41
Ven`` joined
|
|||
| Geth | MoarVM: 7d00853f1d | (Samantha McVey)++ | .appveyor.yml Turn off the over verbose Appveyor nqp tests |
21:18 | |
|
22:46
AlexDaniel joined
22:48
travis-ci joined
|
|||
| travis-ci | MoarVM build failed. Samantha McVey 'Turn off the over verbose Appveyor nqp tests' | 22:48 | |
| travis-ci.org/MoarVM/MoarVM/builds/278785860 github.com/MoarVM/MoarVM/compare/6...00853f1dd4 | |||
|
22:48
travis-ci left
|
|||