🦋 Welcome to the IRC channel of the core developers of the Raku Programming Language (raku.org #rakulang). This channel is logged for the purpose of history keeping about its development | evalbot usage: 'm: say 3;' or /msg camelia m: ... | Logs available at irclogs.raku.org/raku-dev/live.html | For MoarVM see #moarvm
Set by lizmat on 8 June 2022.
00:04 jgaz joined 00:07 AlexDaniel left 01:02 tellable6 left, tellable6 joined 01:07 tellable6 left, tellable6 joined 05:28 committable6 left, bisectable6 left 05:30 bisectable6 joined 05:31 committable6 joined 07:33 sena_kun joined 08:02 jgaz left 08:50 finanalyst joined
nemokosch I wrote at #raku-steering-council for the lack of a better option known to me - apparently the only people who "showed interest" were some sort of alt-right trolls 09:28
lizmat FWIW, I have no idea where your comment was deleted, what it was about or who deleted it 09:30
09:30 sena_kun left
nemokosch I posted a screenshot on # raku-steering-council, couldn't find the original one but it was highly similar. Maybe those who can delete it can see what got deleted anyway? I don't know 09:32
lizmat the screen shots don't make clear which issue this is about ? 09:35
nemokosch that was the comment that got deleted right away, Github fortunately doesn't retroactively remove comments without a full site load 09:37
lizmat what is the URL of the issue ?
nemokosch github.com/Raku/doc-website/issues/304
lizmat ok, I can see who did the deletion 09:41
left a comment 09:43
nemokosch do I feel attacked when I get silently blocked from a discussion that I'm somewhat involved in, when I actually make the point that the "solo creator" shouldn't be drained by vague objections to the architecture? frankly, I do 09:44
09:49 |Tux| left
the irony is that not for the first time, these objections are raised by people who I don't think have actually tried to use the system so I think there would be some merit in saying "I have tried to use it and it seems workable"... 09:55
09:56 |Tux| joined, Tux__ joined 09:59 Tux__ left
lizmat notable6: weekly 11:10
notable6 lizmat, 1 note: 2023-10-01T13:21:27Z <AlexDaniel>: test note
lizmat notable6: weekly reset
notable6 lizmat, Moved existing notes to “weekly_2023-10-02T11:10:54Z”
lizmat and yet another Rakudo Weekly News hits the Net: rakudoweekly.blog/2023/10/02/2023-...n-4-weeks/ 11:17
Geth rakudo/main: c814416715 | (Elizabeth Mattijsen)++ | src/Raku/Actions.nqp
RakuAST: small tweak in term:sym<name> handling
12:08 AlexDaniel joined
[Coke] just had a failure installing Text::CSV on 2023.09. 12:26
tellable6 2023-10-01T20:40:54Z #raku-dev <finanalyst> [Coke] would you have another look at the new search PR for doc-website. Maybe a conf call to discuss whats needed to complete?
[Coke] zef version 0.19.1 12:27
12:36 AlexDaniel left, AlexDaniel joined
ugexe what error? 12:39
hmmm, i can't even install zef on 2023.09 using rakubrew 12:41
raku -e 'say CompUnit::RepositoryRegistry.repository-for-name("site").can-install()' 12:43
Does that one-liner return True for anyone with a 2023.09 built from scratch (not upgraded from a previous version)? 12:44
AlexDaniel mmm yes
tellable6 hey AlexDaniel, you have a message: gist.github.com/40dde62cff9c842dc7...e3486bb49b 12:45
AlexDaniel I get True on 2023.09
installed with rakubrew
pretty sure I wiped everything before installing it
ugexe i see, thanks. im just gonna nuke it and try again i suppose, because nothing changed in the release to have caused that 12:46
[Coke]: Text::CSV fails its test for me as well 12:50
(nuking 2023.09 and rebuilding it fixed my other problem apparently)
[Coke]: lizmat: I think the files/ directory of Text::CSV may not have gotten packaged when it was uploaded to fez 12:57
which the tests use
lizmat [Tux] ^^
ugexe i dunno if/how to get fez to include arbitrary directories, but moving files/ to i.e. t/files would work 12:59
[Coke] the Text::CSV is a blocker for me for $dayjob. I just tried to update everything so I can re-run my data collection, and am blocked - will have to downgrade something (maybe I can install a slightly older Text::CSV) 13:47
:ver<0.014> installed 13:53
ugexe you could just --/test 14:10
it is only failing because the .tar.gz doesn't include a csv file the tests expect to exist
[Coke] I always feel weird about that, but yah 14:16
|Tux| just executed what I was instructed to do. So I move files to t/files, up the version and push and fe upload? 15:39
[Coke] Try to install it locally before pushing, I guess? 16:21
ugexe you'd have to update any references to the file in there as well 16:25
which i assume is just one
Geth Benchmark: rcmlz++ created pull request #1:
Adding more statistics
tbrowder__ ugexe: i’m a dummy, i didn’t think i was actually “using” lib but i’ve run into that before. i apologize for NOT 16:42
paying attn to yr advice
16:48 coleman left 16:49 coleman joined 16:50 coleman left, coleman joined 18:04 MasterDuke joined 18:07 sena_kun joined 18:17 sena_kun left 18:18 sena_kun joined
lizmat ab5tract: just found another eh, wat? in RakuAST that *may* be easy to fix and possibly fix a lot of tests 18:47
nemokosch tbrowder: do you still have your modifications for File::Find somewhere? I think there was a PR 18:48
lizmat ab5tract: ah, no, it looks like a deparsing issue only 18:49
MasterDuke lizmat: didn't you add some sort of rakuast resolve-lexical-compile-time-value function recently? 19:53
20:02 Altai-man joined 20:04 sena_kun left 20:39 NemokoschKiwi joined 20:49 NemokoschKiwi left
Geth Benchmark/main: 79f38d778f | (Márton Polgár)++ (committed using GitHub Web editor) | .github/workflows/test.yml
Attempt to fix testing for Windows as well
nemokosch Apparently it did the trick 21:33
21:49 Altai-man left
ugexe `if: runner.os != 'Windows'` is the more idiomatic way to do that fwiw 22:12
i guess if you want to do a command conditionally in a single step though that wouldn't work 22:17
nemokosch I don't know. Apparently if on steps is relatively recent overall 22:26
ugexe I've been talking to tony about trying to release the next major version of zef (zef 3 - don't ask about 1 or 2) around the time of the next rakudo release. We are planning on breaking backwards compatibility where it makes sense regardless of the ecosystem consequences. The current iteration of zef would continue to live, and would be updated/used to nag people off doing things that will be 22:27
broken in zef 3. Eventually some critical mass of of distributions would work with zef 3, at which point it could be probably be merged into the main branch and become the new zef.
next raku release rather
nemokosch That may as well be great news, depending on what it will affect exactly 😛 22:29
ugexe anything that seems wrong is probably on the table. my goal will mostly be continue the design goal of workflows being parallizable and how that information gets displayed to the user, as well as security considerations 22:36
22:37 finanalyst left
ugexe Build.pm? gone or changed. Module name queries? gone. installing to multiple repos in a single command (lol) gone 22:37
thats a few of the things
vrurg_ What will replace Build.pm? 22:40
ugexe there is Distribution::Builder::MakeFromJSON for instance, but I'll probably write a builder module for doing what Build.pm does (probably just run it even). The main problem there being Build.pm isnt listed in META6.json so there is no way to know if a build step needs to occur without extracting the distribution 22:41
(so users would just add like "builder":"Distribution::Builder::DeprecatedBuilder" to meta6.json, or we inject that key in the ecosystem automatically temprialrly) 22:42
vrurg_ I like it! 22:44
lizmat MasterDuke: .literalize
nemokosch sounds reasonable 23:04
circling back to github actions: isn't there something like premade actions? I think most of Raku repos that are really just distributions in the ecosystem could mostly use the same action file 23:05
ugexe users should probably be using Raku/setup-raku@v1 with zef install . 23:10
just running tests isn't really sufficient
installing it ensures it, well, installs. notably that all the modules it provides can actually compile, which the tests may not exercise for a given module 23:12
nemokosch that's also a fair point but where I'm coming from is like two steps behind from that 23:16
namely that currently there are dozens of various kind of wrong workflows all around
even having the same one wrong workflow would be a step forward, which could then become the one right workflow 23:17
ugexe yeah ci is famous for cargo culted content
gist.github.com/ugexe/8d80f80e2955...c295768476 23:24
this is what i'd use if i was adding a github action to a basic distribution
i created a distribution in the early days to teach people how to do a basic travis-ci config - github.com/ugexe/P6TCI - maybe someone should create one for github actions 23:25
nemokosch do you not think this workflow would be affected by the mysterious TAP error on Windows? 23:26
ugexe If tap::harness is installed by default in those actions (I'm not sure that they are) then i'd add --/tap-harness regardless of which OS 23:29
if its not installed then just don't install it 23:30
nemokosch well, definitely worth a try 23:32
Geth Benchmark/main: 0da63ec72b | (Márton Polgár)++ (committed using GitHub Web editor) | .github/workflows/test.yml
Polished workflow

With a little inspiration from ugexe (jobs part verbatim copied from a gist)
nemokosch yes, this is good. Thanks 23:35
It would make sense to replace a bunch of broken CI's but it's way too much night for me to start with that 23:36
ugexe the only issue that might happen is if it has a dependency that it might go over the windows long path name limit. any distribution without any dependencies (like Benchmark) wouldnt have that problem 23:39
nemokosch I'd say getting an error visibly because of that isn't so bad still 23:41
better than getting an error because of a faulty workflow or a nonreproducible error somewhere in the toolchain
ugexe the error won't give you any indication of what the problem is though
so that part is annoying
nemokosch doesn't it produce some logs? 23:42
ugexe it'll say it can't find some module during testing, which will be due to a precompilation file not getting created or being readable due to path length 23:44
nemokosch ehh 23:48
23:52 guifa left 23:53 guifa joined