04:38 ShimmerFairy left 05:53 hurufu joined 06:12 hurufu left 06:32 hurufu joined 07:01 samebchase left 07:41 ShimmerFairy joined 09:02 timo left
lizmat wow 09:14
ab5tract yeah I clearly made a mistake when I applied the tests committed tag. I wonder which ticket I intended that for... 11:07
ugexe: I've added some comments about ==>> .... I can respect being told to keep it to the ecosystem, but the tl;dr is I think we could get more bang for our NYI-reserved-word buck 11:33
12:06 Geth joined 12:12 lizmat_ left, lizmat joined 13:44 timo joined 14:49 hurufu left 15:04 finanalyst joined
Geth rakudo: w3lld1++ created pull request #6464:
I rename RakuAST currying to priming
15:40
ugexe ab5tract: yeah i did what was in the old design docs, but it might not make sense to do them that way anymore since ==> doesnt follow the spec either 15:57
i.e. ==> already appends, it doesn't clobber as described in those docs
16:10 apogee_ntv left
ab5tract Sweet :D 16:12
a bit ago I started on a user-space "Feed 2.0" but I didn't get it off the ground 16:13
It would have been a lot easier to approach it from the existing RakuAST class 16:14
I think ==>> has the chance to become a mainstay, it's precedence is loose enough that things like my `11..99 Z 111..999 ==>> ...` example would likely draw in people who've never "fed" before 16:17
16:22 apogee_ntv joined 16:23 finanalyst left 16:24 finanalyst joined 16:48 finanalyst left 16:49 finanalyst joined
ab5tract I feel like we could do interesting things in the optimizer, too. 17:16
ugexe: in case you aren't tired of hearing it yet: thank you for all these PRs :) 17:17
Geth rakudo/main: fec98e4ce6 | w3lld1++ (committed by ab5tract) | 9 files
I rename RakuAST currying to priming

I update the internal methods, thunk classes, fields, and comments to use the prime/priming terminology requested in #6199. I also update the RakuAST TODO entry for the renamed PrimeThunk class.
17:19
ab5tract the curry -> prime change is really significantly more readable in my mind. 17:31
ugexe The one lta part of it is primes can mean many things 17:48
Does IMPL-PRIMES implement something to do with is-prime for example
ab5tract I guerss I can see that, but somehow this whole prime code tangle feels a lot more readable to me 17:49
I guess I didn't know before merging this just how stuck my mind was on the word "curry" 17:50
especially since somewhere floating around was "but don't forget, it's not _really_ a curry"
ugexe Seems like that should be in the PR somewhere 17:51
ab5tract sorry, what was missing from the PR? 17:52
ugexe The modification of whatever comment you were alluding to
ab5tract Yeah, in retrospect, I should have sent it back for a better commit message, or amended it myself
ugexe: oh, I meant the fact that this prime vs curry thing comes up quite a bit. so while my brain is trying to parse all these Curry classes and methods, somewhere in my brain is also floating around both "what exactly is a curry again?" and "what exactly was the difference between academic currying and what we do in Raku again?" 17:54
ugexe seems like some of the rakuast tests still referencing currying as well 17:56
ab5tract yeah, I kept the issue open and noted that 17:58
ugexe yeah the rename makes sense to me. i just wish the plural primes wasn't ambig
ab5tract agreed
ugexe: I just had the pleasure of doing my first `make` of rakudo with RAKUDO_RAKUAST=1 is set 18:13
s/with/where/ 18:14
ugexe cool, hopefully it wont be long before its the default 18:18
im not sure its worth having a period where we build the core with rakuast but its not turned on by default. that was in jnthns original plan, but i think that was just so functionality wouldnt regress while the test and spectest was finished. since the test and spectest already work, and so does most of the ecosystem, i dont think it makes much sense 18:21
ab5tract re: the plural primes vs prime, at least '"number" prime' should not be sneaking into AST classes any time soon. and likewise the primes-as-a-language-construct should rightfully appear rarely outside of Language Construct Land 18:22
ugexe: do you have visibility on how much is still in the way of a blin run? 18:27
ugexe there are the serialization issues still which last count was like 50 distributions. there was also like 15-20 that failed due to nativehelpers::blob which i've since (hopefully) fixed, so maybe those 15-20 will get further and expose some other issues. everything else though i believe i've fixed 18:29
even after a clean Blin we might want to do something like try to have people install stuff that Blin doesn't test. for example Red (which i've already done, but just as an example) 18:31
ab5tract right. I wonder if it would be fun to rig up a distributed smoke harness, like back in the day 18:33
18:49 hurufu joined
[Coke] .seen geth 18:49
tellable6 [Coke], I saw geth 2026-07-19T17:20:00Z in #raku-dev: <Geth> ¦ rakudo/main: ␂review: github.com/rakudo/rakudo/commit/fec98e4ce6
[Coke] Is geth now reporting only on PRs and main? 18:51
Waiting for the latest commit to compile before kicking off blin-ast run 18:52
lizmat [Coke]: what are you missing in Geth reporting ? 18:54
releasable6 Next release in ≈4 days and ≈23 hours. There are no known blockers. Please log your changes in the ChangeLog: github.com/rakudo/rakudo/wiki/ChangeLog-Draft 19:00
[Coke] should 467694501b8d5a785767feeabbbdb64922d50f6d have shown up? 19:48
ab5tract [Coke]: I think not, actually 20:01
should be good to go in after the release though 20:02
hmm... actually. on second thought, it's proven to be able to compile rakudo-m, so maybe it is actually time?
ugexe I don’t know how well optimizations work across frontends 20:03
ab5tract fair
[Coke] ab5tract: ? 20:05
I was saying my commit should have shown up in geth, not main 20:06
it's just to do a blin run with AST enabled
ab5tract oh, I thought you were asking whether it should have shown up in the list of new commits to include in blin (which it is, just not in the way I first understood) 20:07
[Coke] there's 3 or 4 of those mini branches lying around for the previous ones. 20:13
ast blin run is going. 20:21
3 fails so far, all Object does not exist in serialization context 20:28
I'll push updates here: github.com/coke/raku-ast-blin-results
20:35 finanalyst left 20:36 finanalyst joined 20:53 guifa_ left 20:54 guifa joined
[Coke] [CompUnit::Util] steal-globalish can only be called at BEGIN time 21:14
ugexe That module uses $*W 21:16
I don’t expect it to ever work on rakuast 21:18
[Coke] [Concurrent::Stack] No exception handler located for catch
[Control::Bail] This type (Failure) does not support associative operations 21:19
ab5tract hmm, that sounds kind of curious. a Failure that was thrown about, rather than thrown itself 21:20
21:22 finanalyst left