|
06:37
andinus2 left
06:38
andinus joined
09:11
sena_kun joined
09:27
sena_kun left
09:28
sena_kun joined
09:33
sena_kun left
|
|||
| Geth | roast: 2795530cf3 | (Elizabeth Mattijsen)++ | S02-types/num.t Group #2434 tests together MasterDuke++ for the nudge |
09:39 | |
|
10:20
andinus left
|
|||
| timo | whee, rakudo-2024.09-1 on arm64 had a rebuild attempt and this time it worked \o/ | 10:49 | |
| lizmat | nice! | 10:50 | |
| Geth | roast: 42efdb75be | (Elizabeth Mattijsen)++ | S06-other/main.t Add test for #2445 |
||
| timo | i wonder why buildd.debian.org/status/logs.php?...=2024.09-1 | 10:51 | |
| er | |||
| ignore the text before the link | |||
| Geth | roast: 6b4e89cc4d | (Elizabeth Mattijsen)++ | S03-operators/assign.t Add test for #2490 |
11:50 | |
| roast: 4f4fe9090a | (Elizabeth Mattijsen)++ | S32-exceptions/misc2.t Add test for #2492 |
12:01 | ||
|
12:28
sena_kun joined
|
|||
| Geth | rakudo/main: 19a32df007 | (Elizabeth Mattijsen)++ | t/12-rakuast/xx-fixed-in-rakuast.rakutest Add tests for #2541 |
12:34 | |
|
12:39
andinus joined
|
|||
| lizmat | greppable6: Stringy | 12:55 | |
| greppable6 | lizmat, 91 lines, 41 modules: gist.github.com/fd34065289f35354b0...bbf16eb6b3 | ||
|
13:08
Voldenet_ joined
13:09
Voldenet left,
Voldenet_ is now known as Voldenet
16:53
[Coke] joined
|
|||
| lizmat | so I had a bit of an epiphany re the "handles" trait | 17:31 | |
| this is currently handled by generating a method and adding this to the class, which basically means an extra level of indirection | |||
| I was thinking we could be smarter in new-disp | 17:32 | ||
| 1. add a %handles hash in the ClassHOW | 17:33 | ||
| 2. let the "handles" trait add "method-name" => '$!attribute-name' pairs to that hash | |||
| 3. let method dispatch first look at this %handles hash | 17:34 | ||
| 4. if the method name exists in that hash, call ^find_method on the associated attribute, and dispatch to that | 17:36 | ||
| 5. if not, use current dispatch | |||
| not only would that save in runtime for handled methods, it would also save in lookup as the current situation really means two full dispatches for handled methods | 17:38 | ||
| nine ab5tract vrurg does that feel like a viable approach ? | |||
| greppable6: handles | |||
| greppable6 | lizmat, 788 lines, 188 modules: gist.github.com/782fc3b1c54ee5d13f...80ccd68e2b | ||
| ab5tract | interesting idea | 17:39 | |
| For #3, there should probably be a flag that is added to any class that uses handles, so that the hash is not checked otherwise | 17:40 | ||
| lizmat | nqp::atkey on empty hashes is close to a noop | 17:41 | |
| m: use nqp; my $h := nqp::hash; nqp::existskey($h,"a") for ^100000000; say now - ENTER now | 17:44 | ||
| camelia | 0.561074376 | ||
| lizmat | m: use nqp; my $h := nqp::hash; Nil for ^100000000; say now - ENTER now | ||
| camelia | 0.403713711 | ||
| lizmat | m: use v6.*; use nqp; my $h := nqp::hash; nqp::existskey($h,"a") for ^100000000; say (nano - ENTER nano) / 100000000 | 17:46 | |
| camelia | 5.71919547 | ||
| lizmat | m: use v6.*; use nqp; my $h := nqp::hash; Nil for ^100000000; say (nano - ENTER nano) / 100000000 | ||
| camelia | 4.12310636 | ||
| Geth | rakudo/main: 7d080edd9e | (Elizabeth Mattijsen)++ | t/12-rakuast/xx-fixed-in-rakuast.rakutest Add test for #2581 |
17:56 | |
| roast: 6dca656e19 | (Elizabeth Mattijsen)++ | S03-smartmatch/capture-signature.t Add test for #2596 |
18:12 | ||
|
21:44
sena_kun left
|
|||
| [Coke] | timo: github.com/rakudo/rakudo/issues/1257 ? | 21:58 | |
| timo | -v | 22:03 | |
| [Coke] | Do we want to create a "6.e" milestone for rakudo/rakudo bugs? | 22:05 | |
| With the thought that anything with that milestone should be fixed by 6.e? (rather than a tag, I find milestones good for grouping things by date or release) | 22:06 | ||
| timo | is that text related to that issue? i'm confused | ||
| [Coke] | nope | 22:07 | |
| I linked that issue just because it was on debian | |||
| timo | oh | 22:10 | |
| did you check the last few comments on that issue? it's only been kept open because there's no automated tests for BE systems? | 22:13 | ||
| [Coke] | Nope! | 22:19 | |
| Geth | rakudo/coke/justyet: 83ea6e1aca | (Will Coleda)++ | 2 files Remove playful "just yet" in error messages. It's not explicitly checked for in test/spectest |
23:14 | |
| rakudo: coke++ created pull request #5656: Remove playful "just yet" in error messages. |
23:15 | ||
| timo creates a pull request to replace it with something even more playful | 23:26 | ||
| [Coke] | O_.; | 23:27 | |