Xliff \o 00:53
jdv79 on the topic of more org/bureaucracy/etc - how do other volunteer efforts do it? i don't remember, though i barely paid attention, there being a lot of ceremony around perl's dev efforts 01:43
iirc there's a pumpkin, others may or may not apply for grants, etc. is there substantionally more in terms of core dev? 01:44
one more random thought - if the first person to do some new role doesn't need a dayjob it'd be pretty hard for someone else (who does have a dayjob) to take it up, no? 01:56
ShimmerFairy Sure, but as a counterpoint what if getting the job started demands more work than keeping it going? 03:33
jdv79 Sure, but i wasn't making a point; just a thought - there's no point. 03:36
[Coke] The grants are really separate from any kind of hierarchy. IME, they're just interested people who wanted to be able to spend more time on the project. 04:49
zzz
lizmat Files=1306, Tests=111378, 215 wallclock secs (28.96 usr 8.34 sys + 3017.68 cusr 279.19 csys = 3334.17 CPU) 09:14
Geth rakudo/rakuast: 825bac4800 | (Jonathan Worthington)++ | 5 files
Add RakuAST for (...) and [...] circumfixes
 10:31
Geth ¦ rakudo: tbrowder self-assigned Add 'truncate-to' as alias for DateTime's 'truncated-to' method github.com/rakudo/rakudo/issues/3750 11:21
Geth rakudo/rakuast: 0139dac152 | (Jonathan Worthington)++ | 2 files
Add RakuAST for [...] and {...} postcircumfixes
 11:46
lizmat jnthn: ^^ copy-pasta in commit mess ? 11:47
jnthn ? 11:49
Looks correct to me 11:50
lizmat aaah... dyslexics untie! 11:59
jnthn It was quite similar to the previous one, I guess... :) 12:02
Up to two sanity tests compiling/passing via RakuAST
moritz \o/ 12:15
timotimo woohoo 13:42
Geth nqp: 30b1d064da | (Jonathan Worthington)++ | 3 files
Allow more interesting constants in NQP

Restrictions are largely like BEGIN blocks: you can't see outer symbols that aren't types.		 13:43
Geth rakudo/rakuast: ecd55ec016 | (Jonathan Worthington)++ | 2 files
Add a range of ops

Including some of the logical ones. Use improved `constant` support in NQP to have a hash we can look up QAST mappings in for that case of ops that need more than just the usual lookup-based compilation.		 13:44
ShimmerFairy I love seeing rakudo slowly hooking into RakuAST more just through the commits. 13:45
Altai-man_ releasable6, status 13:46
releasable6 Altai-man_, Next release in ≈10 days and ≈5 hours. There are no known blockers. 143 out of 266 commits logged (⚠ 4 warnings)
tellable6 2020-06-09T08:36:21Z #raku-dev <patrickb> Altai-man: This is the raku.org revamp by CIAvash: github.com/Raku/problem-solving/issues/158
releasable6 Altai-man_, Details: gist.github.com/5e72c9c06518402307...0ce8aff986
jdv79 whatever happened to PEA? that's the last large perf item i remember. 13:53
jnthn I got distracted by the rename, then with buying an apartment, then when the dispatch thing came up and I started thinking about it, I realized at least one thing I'd need to do in it would give me a solution to one of the PEA problems, so it made sense to return to it once dispatch stuff is done. 13:59
Geth rakudo/rakuast: 1c546f24f3 | (Jonathan Worthington)++ | 4 files
RakuAST handling of chaining operators

Whether something is chaining is a property known by the parser, thus at the RakuAST level it seems reasonable to have that modeled as a separate element type. We can't just have a list of chaining ops in the AST, as we might end up with user-defined chaining ops also. The alternatives are to pass properties into the RakuAST nodes (but which ones?) or to interrogate metadata on the op itself (but that's rather painful for bootstrapping).		 14:18
jdv79 ah, cool 14:21
Geth rakudo/rakuast: 6602064aab | (Jonathan Worthington)++ | 3 files
Wire up very basic RakuAST parameter handling

Just wiring up the grammar/actions to what is implemented in the AST so far, which is really just positional parameters without any regard to type.		 15:08
Geth rakudo/rakuast: 9800adc5e6 | (Jonathan Worthington)++ | 5 files
Make PointyBlock a kind of Block

As well as avoiding some code duplication, this also deals with the various places that accept both a block and a pointy block, meaning that `while foo() -> $x { }` now compiles with RakuAST also.		 16:28
Geth rakudo/rakuast: eb7f1cfbc5 | (Jonathan Worthington)++ | 4 files
Add basic RakuAST `for` loop support

So far doesn't know the optimized forms nor labels, but otherwise does the right thing (so long as there's a pointy block, at least; blocks with implicit topics aren't implemented yet).		 17:00
patrickb o/ 17:08
Currently the AzureCI configs in raku, nqp and moar are meant to be identical. This "keep them the same" approach stemed from the assumption that identical tests would be run for all three projects just with different ways to reference the respective other repos. 17:09
But that assumption is not true for MoarVM for which we want to test a set of different compilers and library combinations. 17:10
patrickb Now I'm unsure whether I should continue putting repo specific special cases into the one big azure ci config or just give up on that idea and accept that the configs are different. 17:11
From a maintainance point of view it's difficult to argue for one of the two approaches. 17:12
github.com/MoarVM/MoarVM/blob/mast...s.yml#L100 <- That's what we currently have. Lines 100-146 are exlusive for MoarVM. 17:14
Any opinions on this? 17:15
I'll be afk for a bit (I need to get some groceries) but will backlog in a bit. 17:16
Geth rakudo/rakuast: da2d3a0d01 | (Jonathan Worthington)++ | 4 files
Add RakuAST for without statements
 17:17
timotimo is that rakuast for "without" statements, or rakuast "for" but without statements inside? :D 17:19
(looking at the code makes it clear) 17:20
jnthn haha :D 17:23
[Coke] I admit I may not be thinking of full context, but I would treat the 3 repos separately. 17:30
what does .kok stand for? 17:31
jnthn keyword ok, I assume :) 17:34
(It's straight out of the current grammar, and I think it's a Larryism :)) 17:36
nine .kok reminds me very much of perl's internals with SvIOK, SvROK and friends... 18:06
MasterDuke makes me think of starcraft 18:08
softmoth Hi, friends. I have a GitHub / Travis request. Yesterday I pushed a button on travis-ci.org (.org is important) requesting that it be authorized by the github Raku org. Someone saw that request and approved it. They should not have, since I was in error (we are using travis-ci.COM, not .ORG). I'm not sure where to report this issue. Who tends to github.com/Raku settings, and where do they prefer issues be filed? 18:19
Altai-man_ softmoth, I can deny it. Should I? 18:21
softmoth I think it's best. Before, this:(travis-ci.org/github/Raku/rakudoc) gave a LTA error saying the repo wasn't available. That's a Travis issue, IMO. But now it gives me an option of activating the repo. But it's already active over on travis-ci.com/Raku/rakudoc, so I think it's best not to allow duplicates. 18:23
Altai-man_ Currently: No private access 18:24
softmoth, ^ should be okay now?
softmoth I'd argue that we should have gone with travis-ci.org/ for github/Raku/* in the first place. But I've only just yesterday learned that there's a distinction, so I could be a victim of severe ignorance here.
Altai-man_, hang on, it's still showing but I noticed sometimes I have to log out & back in to travis to see updates. 18:26
Hmm, it still shows the Activate button now. Ah, but yes, it does fail to activate if I press it. So I think it's okay now. 18:30
Travis has some LTA error handling here, but that's okay. 18:31
Thanks, Altai-man++
Altai-man_ softmoth, you are welcome. Answering to a broader question - you can ping me in such situations, I think. When my connection often dies here, I can appear here as `sena_kun` as well, either works fine. 18:33
softmoth ☺ 18:35
Geth rakudo: 272c46369a | (Elizabeth Mattijsen)++ | src/core.c/Match.pm6
Make Match!MATCH-CAPTURES about 30% faster

  - possibly visible with regexen with a lot of captures
  - simplify indexing into arrays (native str coerces to int automatically)
  - don't use iterator on native str array, just shift from it
Otherwise, kept the exact flow for now. Puzzling is why a lot of things are allocated (like Arrays instead of native lists). But that's for the next round of improvements.		 19:17
rakudo: ed65d733c6 | (Elizabeth Mattijsen)++ | src/core.c/Match.pm6
Make recursive captures a tiny bit faster

  - add a Match!SUBMATCH method that does not check for setuppedness
  - call that instead of !MATCH inside !MATCH-CAPTURES		 19:36
Geth rakudo: e3fa44ab77 | (Elizabeth Mattijsen)++ | src/core.c/Match.pm6
Make Match!MATCH about 10% faster for repeated lookups

  - use NQPdidMATCH type object as a semaphore, instead of an instantiation
  - use nqp::eqaddr instead of nqp::istype for check		 19:51
Altai-man_ committable6, say DateTime.new.truncated-to; 20:06
committable6 Altai-man_, Seems like you forgot to specify a revision (will use “v6.c” instead of “say”)
Altai-man_, gist.github.com/624edcdbc4f8778fdd...37d8dfc726
Geth rakudo: 7f47db80fa | (Elizabeth Mattijsen)++ | src/core.c/Match.pm6
Use faster .push rather than .append

Not sure why a. Array is being used here, and b. why append. Spectest appears clean on using .push instead, opening up the possibility of using an nqp::list rather than a HLL Array, which would save a *lot* of allocations for captures.		 20:07
lizmat afk& 20:08
Altai-man_ committable6, v6c DateTime.new(now).truncated-to('seconds')
committable6 Altai-man_, ¦v6c (43 commits): «»
Altai-man_ committable6, v6.c say DateTime.new(now).truncated-to('seconds')
jnthn I think I've already been down the List instead of Array path for Match, but I forget what happened...
committable6 Altai-man_, gist.github.com/3b305846a49a563345...bdb18a3467 20:09
lizmat jnthn: do you recall why using .append instead of .push ? 20:16
really afk& 20:20
jnthn lizmat: no, and can't think of why it's doing that 20:22
Geth nqp: b4ac42ac53 | (Patrick Böker)++ | azure-pipelines.yml
Simplify the Azure CI configuration + Several fixes

  - Don't try to have one config for rakudo, nqp and MoarVM. This simplifies
   the config a lot.
  - Fix building of PRs
  - Shorten job names for better readability		 20:30
patrickb There currently is no IRC hook for Azure CI. Is that a blocker for getting rid of Travis / AppVeyor / ... ? 20:32
Geth nqp: 0e84bf73b5 | (Patrick Böker)++ | 3rdparty/nqp-configure
Update nqp-configure

Pulls in a set of Perl6 -> Raku renames and a fix for nmake detection with non english locales.		 20:40
Geth rakudo: fce783a182 | (Patrick Böker)++ | azure-pipelines.yml
Simplify the Azure CI configuration + Several fixes

  - Don't try to have one config for rakudo, nqp and MoarVM. This simplifies
   the config a lot.
  - Fix building of PRs
  - Shorten job names for better readability		 20:44
rakudo: 7bace1bef4 | (Patrick Böker)++ | 3rdparty/nqp-configure
Update nqp-configure

Pulls in a set of Perl6 -> Raku renames and a fix for nmake detection with non english locales.
travis-ci Rakudo build passed. Elizabeth Mattijsen 'Make Match!MATCH about 10% faster for repeated lookups 21:57
travis-ci.org/rakudo/rakudo/builds/696981830 github.com/rakudo/rakudo/compare/e...fa44ab7788
travis-ci Rakudo build passed. Elizabeth Mattijsen 'Use faster .push rather than .append 22:48
travis-ci.org/rakudo/rakudo/builds/696986718 github.com/rakudo/rakudo/compare/e...47db80fa4b
Geth rakudo/rakuast: 4b438ecd14 | (Jonathan Worthington)++ | 4 files
Add if/with/elsif/orwith/else to RakuAST
 22:55
Geth ¦ rakudo: tbrowder self-unassigned Add 'truncate-to' as alias for DateTime's 'truncated-to' method github.com/rakudo/rakudo/issues/3750 23:02

Powered by Cro and the Raku® Programming Language.

Please report any issues / comments / feature requests as an issue on App::Raku::Log.

Thank you!