|
github.com/moarvm/moarvm | IRC logs at colabti.org/irclogger/irclogger_logs/moarvm Set by AlexDaniel on 12 June 2018. |
|||
| MasterDuke | timotimo: fyi, ran the --profile-compile of the rakudo build in gdb, it just exited normally (but no profile actually written) | 00:37 | |
|
02:44
MasterDuke_ joined
02:45
MasterDuke left
03:10
lizmat left
03:27
MasterDuke_ is now known as MasterDuke
|
|||
| Geth | MoarVM: f75abe7c69 | (Samantha McVey)++ | src/strings/uthash.h Fix the grammar of a comment in uthash.h |
03:28 | |
|
03:52
MasterDuke left
06:24
lizmat joined
07:29
[Coke] left,
[Coke] joined
08:08
Geth joined,
synopsebot joined
08:12
lizmat left
08:18
domidumont joined
08:25
domidumont left,
domidumont joined
10:22
domidumont left
12:59
MasterDuke joined
|
|||
| MasterDuke | timotimo: what's the sequence of functions called in moar after normal execution is done and writing out profiling data is starting? | 13:12 | |
| i.e., my --profile-compile of the rakudo build doesn't get to dump_instrumented_profile_data in HLL/Backend.nqp, so i want to add some debug prints in moar to see where it does get to there | 13:14 | ||
| timotimo | have you seen src/vm/moar/HLL/Backend.nqp and its run_profiled method? | 13:15 | |
| MasterDuke | hm, that just sticks self.dump_profile_data into an END phaser? | 13:17 | |
| i guess i can stick a note() in there to see if it gets called... | 13:19 | ||
| timotimo | oh, end phaser, huh? | 13:21 | |
| that's probably what's going wrong | |||
| MasterDuke | how so? i mean, --profile-compile does work for some other test scripts i tried, and it used to work for the rakudo build | 13:23 | |
| timotimo | the core setting profile doesn't have a setting, so perhaps the END_PHASERS symbol doesn't exist at the beginning | 13:28 | |
| MasterDuke | did something change with @END_PHASER in the past 9 months? | 13:31 | |
| because in september 2017 i could profile the build | 13:32 | ||
|
14:03
Voldenet joined,
Voldenet left,
Voldenet joined
14:16
lizmat joined
14:44
domidumont joined
|
|||
| dogbert17 | libuv 1.21.0 is out, github.com/libuv/libuv/blob/v1.x/ChangeLog | 16:44 | |
| samcv | .tell brrt when you get the chance can you review my PR for expr jit templates? github.com/MoarVM/MoarVM/pull/873 | 17:08 | |
| yoleaux | samcv: I'll pass your message to brrt. | ||
|
17:22
evalable6 left
17:23
evalable6 joined
18:10
domidumont left
18:46
lizmat left
19:28
brrt joined
|
|||
| brrt | samcv++ # expr templates yay | 19:28 | |
| yoleaux | 17:08Z <samcv> brrt: when you get the chance can you review my PR for expr jit templates? github.com/MoarVM/MoarVM/pull/873 | ||
|
19:56
zakharyas joined
19:59
MasterDuke left
20:06
zakharyas left
|
|||
| Kaiepi | what does it mean when there's a ! after an expr template name? | 20:07 | |
| samcv | brrt: 57 more! whee | 20:09 | |
| Kaiepi: uh the function puts something in the return register without assistance | 20:11 | ||
| so it writes to the return value register directly | |||
|
20:12
zakharyas joined
|
|||
| brrt | the idea is that without a '!', a template forms an expression (that returns a value); the JIT can decide whether it is necessary to store that to memory | 20:12 | |
| the '!' basically constrains that | |||
| dunno if you know about scheme, but it has the same sort of convention :-) | 20:13 | ||
| samcv | brrt: could you figure out how to turn this into a command line script? github.com/yashi/org-asciidoc | ||
| i don't know much about emacs or lisp | |||
|
20:14
zakharyas left
|
|||
| brrt | you probably could, actually, but i've never quite learned the magic incantations to do so :-) | 20:14 | |
| something about emacs --batch somethingsomething | |||
| Kaiepi | i think i understand | 20:15 | |
| samcv | if we can get them all to asciidoc i should be able to make a nice website that has all the pages in an index or something | 20:16 | |
|
20:18
Kaiepi left,
Kaiepi joined
20:19
zakharyas joined
|
|||
| Kaiepi | there are a few jit templates i think i'd be able to add after samcv's pullreq gets merged | 20:21 | |
| samcv | brrt: well this tool seems to work perfectly even with tables in emacs | 20:22 | |
| not sure how to export it to a file though | |||
| i run org-asciidoc-export-as-asciidoc and then it comes up with another emacs window showing the converted file | 20:23 | ||
| brrt | normally org gives you a window to choose how to export | 20:28 | |
|
20:29
MasterDuke joined
|
|||
| brrt | i'm a bit too tired to figure it out at this point though :-) | 20:35 | |
| and there's also the weird thing about Inline::Perl5 and throwpayloadlexcaller | 20:36 | ||
| throwpayloadlexcaller needs a review anyway, I just don't understand that stuff | |||
| speak y'all later | 20:41 | ||
|
20:41
brrt left
|
|||
| samcv | o/ | 21:02 | |
| Kaiepi | are there any cases where it'd be a bad idea to make a jit template for an op? | 21:44 | |
| also, how do i test templates i've added? just use the ops? | 21:45 | ||
|
21:56
zakharyas left
22:02
MasterDuke left
22:10
Kaiepi left
|
|||
| samcv | this looks pretty decent cry.nu/files/mvmdoc/README.html | 22:10 | |
| sample page with text: cry.nu/files/mvmdoc/jit/overview.html cry.nu/files/mvmdoc/collation.html | 22:11 | ||
|
22:11
Kaiepi joined
|
|||
| samcv | haven't dealt with the markdown files yet though but orgmode and asciidoc ones are working | 22:13 | |
|
22:17
Kaiepi left,
Kaiepi joined
22:18
Kaiepi left
22:19
Kaiepi joined
|
|||
| Geth | MoarVM: f12567907d | (Jonathan Worthington)++ | 8 files Add a negated object literal spesh plugin guard This will be useful for specializing on "not Nil" in the Perl 6 spesh plugin for assignment. |
22:28 | |
| MoarVM: 02bddce853 | (Jonathan Worthington)++ | 2 files JIT the object non-match guard |
22:30 | ||
| MoarVM: b3ce4f6d33 | (Jonathan Worthington)++ | src/spesh/plugin.c Elide object non-match guard in a bunch of ways The "must not be object X" guard can be eliminated by noticing any of: * That we know the value and it's not the one we guard against * That we know the value's type and it's not the same type as the one we guard against * Knowing the concreteness of the value is opposite to the one we guard against |
22:40 | ||
|
22:41
travis-ci joined
|
|||
| travis-ci | MoarVM build errored. Jonathan Worthington 'Add a negated object literal spesh plugin guard | 22:41 | |
| travis-ci.org/MoarVM/MoarVM/builds/395942209 github.com/MoarVM/MoarVM/compare/0...2567907d9d | |||
|
22:41
travis-ci left
|
|||
| jnthn | ah, apt failure | 22:42 | |
|
22:48
MasterDuke joined
|
|||
| MasterDuke | Kaiepi: you know about setting MVM_JIT_LOG=<something> and checking there? | 22:49 | |
| Kaiepi | no | 22:54 | |
|
23:09
travis-ci joined
|
|||
| travis-ci | MoarVM build failed. Jonathan Worthington 'JIT the object non-match guard' | 23:09 | |
| travis-ci.org/MoarVM/MoarVM/builds/395942575 github.com/MoarVM/MoarVM/compare/f...bddce853ad | |||
|
23:09
travis-ci left
23:40
MasterDuke left
23:48
MasterDuke joined
|
|||
| MasterDuke | Kaiepi: if you grep for 'BAIL' you'll see non-jitted ops | 23:49 | |
| so then you can try and jit them and see if they don't cause BAILs afterwards | 23:50 | ||