| jnthn did a blog post to explain the specializer plugin mechanism and how the speedups today were achieved: 6guts.wordpress.com/2018/06/09/fas...r-plugins/ | 00:02 | ||
| timotimo | "it works a few steps" sounds a few odd | 00:08 | |
| MasterDuke | "this eliminating", probably meant "thus"? | ||
| timotimo | And that’s in. So what happens at runtime? | 00:09 | |
| ^- probably wants to be "that's it" | |||
| jnthn | Fixed, thanks :) | 00:10 | |
| MasterDuke | "It was the same case story", probably mean either "case" or "story"? | ||
| jnthn | heck knows what I meant :) | ||
| jnthn decides he meant "story" :) | 00:11 | ||
| jnthn is a really awful proof reader | |||
| I read it like twice and didn't spot these :) | |||
| MasterDuke | "every one of them as a precomputed hash table", "has a"? | ||
| jnthn | Yup | 00:12 | |
| MasterDuke | heh, i'm actually kind of decent at proofing other people's work, but am terrible at my own | ||
| jnthn | Yeah, I think that's true of me also | ||
| timotimo | do we have spesh plugin output in the log yet? and if not, should i look into that? | ||
| jnthn | I do wonder if part of it is that I have pretty awful eyesight, and so whenever I read text I rely a lot on word shapes, which makes spotting typos hard enough as it is, but I also have fairly decent memory, and probably subconciously resolve the shapes into my recollection of the meaning. :) | 00:13 | |
| timotimo: Yes, it's in the spesh stats | |||
| timotimo | ql | 00:14 | |
| MasterDuke | "along the the conditions" | ||
| jnthn | github.com/MoarVM/MoarVM/blob/mast...ump.c#L631 | 00:15 | |
| MasterDuke | well, i know i make the same kind of mistake writing text as programming, where i jump ahead in my thought and skip words | ||
| jnthn | Oh yeah, I do that too! | ||
| MasterDuke | and it's hard for me to spot them even when i go back, because i subconsciously know what i meant | ||
| jnthn | Hands can't keep up :) | ||
| I'd do it with handwriting too...always when I hit a letter in the current word that was the first letter of the next word or so :) | 00:16 | ||
| MasterDuke | one of the reasons i love the perls. they're nice an succinct, so i can type/program closer to the speed i'm thinking | ||
| jnthn | Indeed :) | ||
| timotimo | i liked that blogpost | 00:20 | |
| hoelzro | oooo spesh plugins? neat! | 00:21 | |
| hoelzro reads | |||
| geekosaur | abd then you get someone wondering why .sort didn't automatically detect dates in mdy format and sort as such instead of lexicographically. | ||
| timotimo | month, day, year? uuugggggghhhhhh | 00:22 | |
| jnthn | Middle endian. | ||
| geekosaur | muddle endian | ||
| jnthn | When I first heard that term I thought "haha, that's a cute way to describe that date format", then looked it up and to my horror realized that middle endian has actually existed in software too :) | 00:23 | |
| MasterDuke | jnthn++, very understandable | 00:24 | |
| jnthn | MasterDuke: Cool, that was my aim :) | 00:25 | |
| Thanks timotimo++ and MasterDuke++ for the fixes | |||
| timotimo | <3 | ||
|
00:28
j3nnn1 left
00:37
dct left
|
|||
| jnthn | 'night o/ | 00:41 | |
| timotimo | gnite jnthn | ||
|
01:33
japhb left
|
|||
| timotimo | oh, and yes: using 0.999e0 instead of 0.999 for that code example gets it from about 4.75s down to ~2.5s on my box | 01:41 | |
| (don't have a pre-spesh-plugin measurement) | |||
|
01:46
Kaiepi joined
02:32
Kaiepi left
02:33
Kaiepi joined
|
|||
| Geth | nqp: c9f84d39f5 | 陈梓立++ (committed using GitHub Web editor) | t/nqp/020-return.t Add test to make sure `return -1` works Fix was commited here github.com/perl6/nqp/commit/049f5a...6fa3b6cd2b This will close #463 |
02:39 | |
| synopsebot | NQP#463 [closed]: github.com/perl6/nqp/issues/463 Add test to make sure `return -1` works | ||
|
02:55
travis-ci joined
|
|||
| travis-ci | NQP build failed. 陈梓立 'Add test to make sure `return -1` works | 02:55 | |
| travis-ci.org/perl6/nqp/builds/390001386 github.com/perl6/nqp/compare/5f8cf...f84d39f598 | |||
|
02:55
travis-ci left
02:57
Tison left
02:59
lizmat left
|
|||
| Geth | nqp: cf3cfb9ee2 | 陈梓立++ (committed using GitHub Web editor) | t/nqp/020-return.t Correct plan number |
03:36 | |
|
03:51
travis-ci joined
|
|||
| travis-ci | NQP build passed. 陈梓立 'Correct plan number' | 03:51 | |
| travis-ci.org/perl6/nqp/builds/390009698 github.com/perl6/nqp/compare/c9f84...3cfb9ee27e | |||
|
03:51
travis-ci left
06:00
Geth left
06:01
dalek joined,
ChanServ sets mode: +v dalek,
Geth joined,
ChanServ sets mode: +v Geth,
p6lert joined,
synopsebot_ joined,
ChanServ sets mode: +v synopsebot_
07:06
lizmat joined
07:14
skids left
|
|||
| Geth | rakudo: c940a0ed06 | (Elizabeth Mattijsen)++ | 14 files Change all SET-SELF methods back to private methods - now that we optimize them much better - some cases remain because they're called "from the outside" |
07:32 | |
|
08:18
robertle joined
08:26
travis-ci joined
|
|||
| travis-ci | Rakudo build failed. Elizabeth Mattijsen 'Change all SET-SELF methods back to private methods | 08:26 | |
| travis-ci.org/rakudo/rakudo/builds/390046416 github.com/rakudo/rakudo/compare/c...40a0ed06b5 | |||
|
08:26
travis-ci left
|
|||
| buggable | [travis build above] ☠ All failures are due to: failed make test (6 failures). Across all jobs, only t/05-messages/02-errors.t test file failed. | 08:26 | |
| nine | lizmat++ # doing not just exciting speedups but also the boring cleanups | 08:27 | |
|
08:59
robertle left
09:40
lizmat left
09:42
[TuxCM] left
09:49
lizmat joined
10:41
dogbert2 left
10:50
dogbert2 joined
11:34
SourceBaby joined,
ChanServ sets mode: +v SourceBaby
11:35
Undercover joined,
ChanServ sets mode: +v Undercover
|
|||
| cognominal | jnthn++ # thx. Your blog, here the brand new entry 6guts.wordpress.com/2018/06/09/fas...r-plugins/ and your code are a delight to read. | 12:00 | |
| jnthn | cognominal: Thanks, glad you enjoyed :) | 12:07 | |
|
12:33
robertle joined
|
|||
| cognominal | jnthn, Ironically; the ad displayed in your post page was from a company, automattic, recruiting php developper :( | 12:50 | |
| *developpers | 12:51 | ||
| Geth | rakudo: 23af349103 | (Stefan Seifert)++ | 2 files Initialize attributes of meta model roles when used from Perl 6 The meta model roles are written in NQP. When they are composed into a class written in Perl 6, the list and hash attributes will not automatically get initialized. So do it manually. |
13:38 | |
| rakudo: 7bc6e46fbd | (Stefan Seifert)++ | 11 files Cope with ^method_table and ^submethod_table written in Perl 6 When these meta model methods are written in Perl 6, they will return a Perl 6 map instead of a VMHash. Need to unwrap to be able to iterate in NQP. |
|||
| rakudo: cbb5ee5f96 | (Stefan Seifert)++ | src/Perl6/Metamodel/REPRComposeProtocol.nqp Cope with containerized type objects in REPRComposeProtocol |
|||
| nqp: c9fa8215fe | (Jeremy Studer)++ | src/vm/moar/QAST/QASTOperationsMAST.nqp Remove erroneous decont op in chain compilation Removing entirely superfluous decont operation included when compiling MAST for the chain op. Not only was this decont not doing any good, but in some scenarios it would result in error (see [Github Issue 1886](github.com/rakudo/rakudo/issues/1886)) although it's unclear to me why it's sometimes a useless addition and sometimes an explicit error. |
13:51 | ||
| nqp: ee56eb3f44 | timo++ (committed using GitHub Web editor) | src/vm/moar/QAST/QASTOperationsMAST.nqp Merge pull request #455 from jstuder-gh/chain_rm_decont Rakudo Issue 1886: Remove erroneous decont op in chain compilation |
|||
|
14:09
MasterDuke left,
MasterDuke joined
14:30
travis-ci joined
|
|||
| travis-ci | Rakudo build failed. Stefan Seifert 'Cope with containerized type objects in REPRComposeProtocol' | 14:30 | |
| travis-ci.org/rakudo/rakudo/builds/390118330 github.com/rakudo/rakudo/compare/c...b5ee5f96a0 | |||
|
14:30
travis-ci left
|
|||
| buggable | [travis build above] ☠ All failures are due to: failed make test (6 failures). Across all jobs, only t/05-messages/02-errors.t test file failed. | 14:30 | |
|
14:44
AlexDani` joined
14:45
AlexDaniel left
14:49
AlexDani` is now known as AlexDaniel
15:23
skids joined
15:47
Tison joined
|
|||
| Tison | \o | 15:47 | |
| Geth | nqp: 3aa04e3601 | (Stefan Seifert)++ | 24 files Revert "Use faster nqp::elems instead of + on QAST nodes" This reverts commit d743c0f88d2e748bf91caf3813152bd7601b6627. and unbreaks the spectest. dogbert17++ for finding it |
15:48 | |
| MasterDuke | greppable6: nqp::time_i | ||
| greppable6 | MasterDuke, Found nothing! | ||
| timotimo | greppable6: nqp::time_n | 15:49 | |
| greppable6 | timotimo, 28 lines, 5 modules: gist.github.com/7d66af2c5ca671a00c...3b758261a3 | ||
| MasterDuke | it's only used twice in the rakudo, not at all in nqp | ||
| github.com/rakudo/rakudo/blob/mast...y.pm6#L779 and github.com/rakudo/rakudo/blob/mast...t.pm6#L111 | 15:50 | ||
| seems like it could be safely gotten rid of | 15:52 | ||
|
15:55
Tison left
|
|||
| Geth | nqp/NOT_ACCEPTS: 85832d7f7c | 陈梓立++ (committed using GitHub Web editor) | t/nqp/045-smartmatch.t Add test for !~~ |
16:10 | |
|
18:00
camelia left
18:02
bartolin_ joined,
jsimonet1 joined
18:04
bartolin left,
camelia joined,
mst__ joined,
mst left,
mst__ is now known as mst,
mst left,
mst joined
18:05
ChanServ sets mode: +v camelia
|
|||
| Geth | nqp: efb316e550 | (Daniel Green)++ | src/NQP/Optimizer.nqp Make the NQP optimizer convert +@ to nqp::elems(@) |
18:32 | |
|
18:35
skids left
|
|||
| Geth | rakudo: MasterDuke17++ created pull request #1911: Fix arg type in X::Str::Sprintf::Directives::Count |
18:39 | |
|
19:20
dct joined
19:27
lizmat left
19:49
lizmat joined
|
|||
| AlexDaniel | releasable6: status | 20:14 | |
| releasable6 | AlexDaniel, Next release in ≈6 days and ≈22 hours. 3 blockers. 0 out of 64 commits logged | ||
| AlexDaniel, Details: gist.github.com/64f15fdcd5793845e6...22a9658723 | |||
|
20:30
Kaiepi left,
Kaiepi joined
20:32
ilmari[m] joined
|
|||
| Geth | rakudo: ec5e51c8b7 | (Elizabeth Mattijsen)++ | src/core/Rakudo/Iterator.pm6 Make some more methods private |
20:49 | |
| jnthn | So much privatization :) | 20:59 | |
| lizmat++ | |||
|
21:05
AlexDaniel`` joined,
CIAvash[m] joined
21:21
dct left
21:23
dct joined
|
|||
| Geth | rakudo: 7284a30ff6 | (Elizabeth Mattijsen)++ | src/core/Array.pm6 Privatize some more Array methods |
21:26 | |
|
21:28
Kaiepi left,
Kaiepi joined
|
|||
| timotimo | normally i'd be very much against privatisation | 22:00 | |
| AlexDaniel foo | 22:08 | ||
|
22:09
dct left
|
|||
| MasterDuke bar | 22:18 | ||
|
22:40
travis-ci joined
|
|||
| travis-ci | Rakudo build failed. Elizabeth Mattijsen 'Privatize some more Array methods' | 22:40 | |
| travis-ci.org/rakudo/rakudo/builds/390227533 github.com/rakudo/rakudo/compare/e...84a30ff6cd | |||
|
22:40
travis-ci left
|
|||
| Geth | rakudo: c49bbc934b | (Elizabeth Mattijsen)++ | src/core/List.pm6 Privatize some more List methods |
22:40 | |
| buggable | [travis build above] ☠ All failures are due to: failed make test (6 failures). Across all jobs, only t/05-messages/02-errors.t test file failed. | ||
|
23:28
travis-ci joined
|
|||
| travis-ci | Rakudo build failed. Elizabeth Mattijsen 'Privatize some more List methods' | 23:28 | |
| travis-ci.org/rakudo/rakudo/builds/390242002 github.com/rakudo/rakudo/compare/7...9bbc934b68 | |||
|
23:28
travis-ci left
|
|||
| buggable | [travis build above] ☠ All failures are due to: failed make test (6 failures). Across all jobs, only t/05-messages/02-errors.t test file failed. | 23:28 | |
|
23:46
robertle left
|
|||