|
00:25
MasterDuke joined
|
|||
| MasterDuke | github.com/rakudo/rakudo/blob/main...#L471-L478 seems to be the relevant code | 00:26 | |
|
00:26
librasteve_ left
|
|||
| timo | oh i totally missed that there was a -n there | 00:27 | |
| my rakudo isn't up to latest and in there i get 1 as the output, and when i change `my` to `state` I get the number of lines | 00:30 | ||
| MasterDuke | yeah, state works | 00:31 | |
| timo | oh, i thought it outputs the last line rather than the value in the variable? | ||
| MasterDuke | it outputs the value of .words.elems on the last line, instead of the sum of all of them | ||
| timo | oh, that could be | 00:32 | |
| MasterDuke | i.e., as if `$w` was being declared for each line | ||
| timo | i thought that's the same as without rakuast that you need `state` if you want the value to stick around | 00:39 | |
| MasterDuke | not when "predeclaring" it with BEGIN | 00:40 | |
| timo | that's a little odd though | 00:41 | |
| MasterDuke | though i think `state` was created as a more convenient form of using BEGIN for this kind of thing | 00:43 | |
| timo | not sure state was for exactly this :D | ||
| MasterDuke | heh, this is about %110 of my use-cases for state | 00:46 | |
| timo | well, that makes sense :) | 00:49 | |
| I tend to start with -e and lines().map | |||
| that usually makes it easier to expand to a script that has a "second stage" that would otherwise have to be a huge END block | 00:50 | ||
| but of course -ne and -pe are meant to be useful in a larger pipeline of commands, too | 00:51 | ||
| MasterDuke | i would just grow and grow the END block. it would end up wrapping several times in a full-screen terminal | ||
| once or twice i did end up converting it into an actual script, with multiple lines, indentation, the whole works | 00:52 | ||
| earlier today i wrote `perl -nE 'if (/<(\d+\.\d+)>) { $t += $1 }; END { say $t }' <path/to/file>`, because i was on a computer that didn't have raku installed | 00:54 | ||
| right now that sort of thing is my main use case for perl/raku, usually a bit more complicated though | 00:56 | ||
| so bringing it back to the original example. using `RAKUDO_RAKUAST=1` it works with `state`, but not with `BEGIN`, legacy frontend works with both | 00:58 | ||
|
01:36
MasterDuke left
02:59
ShimmerFairy left
03:00
ShimmerFairy joined
03:29
djinni` left
03:42
djinni` joined
03:58
ntv joined,
apogee_ntv left
03:59
ntv is now known as apogee_ntv
|
|||
| ugexe | MasterDuke: nah i've just been focused on getting things to work, although i do look at optimization stuff if it is blocking me, e.g. github.com/rakudo/rakudo/pull/6125 | 04:32 | |
| tellable6 | ugexe, I'll pass your message to MasterDuke | ||
| ugexe | But note zef doesn’t work under rakuast so I’m not sure we’re that close to a clean enough blin run | 04:36 | |
|
08:42
apogee_ntv left
08:43
apogee_ntv joined
|
|||
| Geth | rakudo/main: 4 commits pushed by (Nick Logan)++, (Elizabeth Mattijsen)++ | 10:00 | |
| lizmat | m: dd sprintf(q/%#o/,-10) # meh | 10:21 | |
| evalable6 | Cannot test a716df8f53c61176dbec515cb187a9751130bde2 (Commit exists, but an executable could not be built for it) | ||
| linkable6 | (2026-05-09) github.com/rakudo/rakudo/commit/a716df8f53 Merge pull request #6170 from ugexe/ugexe/issue-6169 | ||
| lizmat | m: dd sprintf(q/%#o/,-10) # meh | ||
| evalable6 | Cannot test a716df8f53c61176dbec515cb187a9751130bde2 (Commit exists, but an executable could not be built for it) | ||
| lizmat | weird | ||
| lizmat rebuilds from scratch | 10:22 | ||
| intriguing... build fails from scratch | 10:25 | ||
| hmmm... cannot repro :-( | 10:34 | ||
| m: dd sprintf(q/%#o/,-10) # meh | |||
| evalable6 | Cannot test a716df8f53c61176dbec515cb187a9751130bde2 (Commit exists, but an executable could not be built for it) | ||
| linkable6 | (2026-05-09) github.com/rakudo/rakudo/commit/a716df8f53 Merge pull request #6170 from ugexe/ugexe/issue-6169 | ||
| Geth | rakudo/main: 0dcdb5c50e | (Elizabeth Mattijsen)++ | src/core.e/Format.rakumod Make Format.raku subclass safe |
||
| lizmat | originally planned as a part of a bigger commit, checking if it builds now | 10:35 | |
| m: dd sprintf(q/%#o/,-10) # meh | 10:37 | ||
| evalable6 | Cannot test a716df8f53c61176dbec515cb187a9751130bde2 (Commit exists, but an executable could not be built for it) | ||
| linkable6 | (2026-05-09) github.com/rakudo/rakudo/commit/a716df8f53 Merge pull request #6170 from ugexe/ugexe/issue-6169 | ||
| lizmat | anyways, the result is: "0-12" | 10:38 | |
| which is wrong | |||
| m: dd sprintf(q/%#o/,-10) # meh | 10:54 | ||
| evalable6 | "0-12" | ||
| disbot2 | <librasteve> wonders if there's Grammar for sprintf s? | 11:05 | |
| lizmat | yes, there is: the original one lives in NQP, and the new one lives in src/core.e/Formatter/Grammar.rakumod | 11:06 | |
| in NQP: src/HLL/sprintf.nqp | 11:07 | ||
| disbot2 | <librasteve> ++ | 11:15 | |
| lizmat | mind you, I did a lot of work on Formatter in the 2020-2023 years... but then I stopped because it couldn't be really put into production because of precomp issues | 11:17 | |
| I now see there's a bunch of issues that need to be fixed, and am working on that | 11:18 | ||
| (what didn't help was that the spectests for the new implementation were not run by standard spectest) | |||
|
11:43
shareable6 left
11:44
shareable6 joined
|
|||
| [Coke] rechecks stuff in Snyk (after fighting it at $dayjob for weeks) | 13:46 | ||
| 32 issues in MoarVM, including: improper null termination, null derefer, use after free, missing release of memory, potential buffer overflow, double free | 13:56 | ||
| Anyone want access to the org so they can see these directly, anyone want to pick one of those categories and force me to open tickets on moarvm/moarvm? | |||
| afk for a bit. | 14:01 | ||
| (also trying to figure out if it can auto -open tickets on github) | 14:02 | ||
|
14:04
librasteve_ joined
|
|||
| [Coke] | oof. bad timing, forgot about snyk maintenance window. :) | 15:14 | |
| ugexe | i think the current eval issue is asymmetry across multiple parallel frontend | 15:38 | |
| one solution would be implement $*CU in legacy, i.e. it would require legacy Perl6::Actions to also construct a RakuAST::CompUnit and bind $*CU to it. but that seems like a lot | 15:40 | ||
| the other oslution would me to make RakuAST EVAL detect when it's running inside legacy and reach for $*W so the bookkeeping works | |||
| i intend to try moving forward with the later since that is the one is easier to model in my head | 15:43 | ||
| [Coke] | gist.github.com/coke/e765618d76792...53c6576751 - all the snyk issues across various repos. | 15:46 | |
| (inludes a bunch of docker issues which I recall someone was adamant about not upgrading docker images for.) | 15:50 | ||
| Any suggestions for "essential" blin testing updates to this list (additions or removals), please let me know: github.com/Raku/Blin/blob/coke/cle...ntial.json | 16:09 | ||
| I can't remember where all the original suggestions came from. | 16:10 | ||
| huh. and many of them are failing already, it seems? | 16:30 | ||
| found a blin issue - Hash::Ordered is an alwaysfail on blin, because of the dep on Hash::Agnostic. (I think it's having a hard time dealing with a pinned version) | 17:39 | ||
| ... pinned version that happens to not be the latest version, that is. (since it's only testing the latest version of something, same issue as with JSON::Class) | 17:42 | ||
| japhb | Why does someone *not* want to upgrade insecure docker images? Is there a compatibility issue? Were they just not wanting to upgrade *major* versions, but revisions would be OK? | 18:04 | |
| lizmat | ugexe: that sounds like the correct way forward, so that we can drop that code later when RakuAST is the default | 18:09 | |
| [Coke]: Hash::Ordered pins as: use Hash::Agnostic:ver<0.0.17+>:auth<zef:lizmat>; | 18:11 | ||
| and Hash::Agnostic is at 0.0.19 so how would that be an issue? what is the exact error ? | 18:12 | ||
| [Coke] | because blin is broken. | 18:13 | |
| blin: HO deps on HA. HA is at 0.019 - tested, works. now let's test HO - oh, I can't find the HA module. die. | |||
|
18:15
bisectable6 left,
unicodable6 left,
quotable6 left,
bloatable6__ left,
benchable6 left,
releasable6 left,
coverable6 left,
committable6 left,
nativecallable6 left,
tellable6 left,
huggable6 left,
shareable6 left,
linkable6 left,
notable6 left,
sourceable6 left,
greppable6 left,
evalable6 left
18:18
linkable6 joined,
bloatable6 joined,
notable6 joined,
greppable6 joined,
sourceable6 joined,
bisectable6 joined
18:19
nativecallable6 joined,
evalable6 joined,
shareable6 joined,
benchable6 joined,
committable6 joined,
releasable6 joined,
quotable6 joined
18:20
huggable6 joined,
unicodable6 joined,
coverable6 joined,
tellable6 joined
|
|||
| Geth | rakudo: ugexe++ created pull request #6176: Fix .assuming dep leak when consumer is precompiled with legacy frontend |
19:07 | |
| [Coke] | patrickb: I don't see how to open issues on git.sr.ht/~patrickb/Terminal-API | 19:19 | |
| the test works if I run it by hand. in blin, the get-config call in the basic test is failing | |||
| patrickb | [Coke] Yeah, the SourceHut UI currently has one major flaw. It doesn't provide uplinks from the components up to the project hub. | 19:21 | |
| You have to remove the "git." in the URL. That will take you to the project which has links to all the components, including the tickets. | 19:22 | ||
| [Coke] I didn't look into it, but it doesn't surprise me that the tests fail in blin. The module tries to interact with the terminal the process is attached to. It's pretty probable, that on blin there is no usable terminal to interact with. | 19:24 | ||
| ugexe | MasterDuke: it looks like the begin and end phasers are getting wrapped by the loop | ||
| tellable6 | ugexe, I'll pass your message to MasterDuke | ||
| [Coke] | can you skip all tests (and warn) if no terminal? | 19:52 | |
| ugexe | anything actually using it would need to do that as well | 19:54 | |
|
21:06
rakkable__ left,
rakkable joined
|
|||
| [Coke] | confirming, this github.com/CurtTilmes/raku-libcurl...on#L12-L24 is invalid, yes? | 21:10 | |
| huh. zef-deps seems fine with it. | 21:12 | ||
| blin does not, however. | 21:13 | ||
| SSL is a dependency of one of the "Essential" modules, but it's no longer maintained, and doesn't pass tests on blin | 21:27 | ||
| Geth | rakudo/main: 35eb4abbb8 | (Nick Logan)++ | azure-pipelines.yml Add CI job for RAKUDO_RAKUAST=1 built rakudo to run RAKUDO_RAKUAST=1 make spectest Add a Lin_Rakuast_Spectest job to the Extra_Moar_Stability_Tests stage that mirrors Lin_Spesh_Nodelay but pulls the rakuast build artifacts (MVM_rakuast_Lin, MVM_sourcecode_built_rakuast_Lin) and sets RAKUDO_RAKUAST=1 before running m-spectest. |
23:10 | |
| rakudo/main: d0d1b716be | (Nick Logan)++ (committed using GitHub Web editor) | azure-pipelines.yml Merge pull request #6174 from ugexe/ugexe/rakuast-ci-spectest Add CI job for RAKUDO_RAKUAST=1 built rakudo to run RAKUDO_RAKUAST=1 make spectest |
|||
| japhb | ugexe++ # Key CI job | 23:58 | |