[Coke] [Inline::Perl5] .pm file extension in raku library path seen at: 00:15
[Coke] so I ran "rakudo-inline-perl5" step of Ake. it succeeded. Then I ran "rakudo-stress" and the first thing it complains about is that Inline::Perl5 isn't installed 00:16
did it by hand, seems OK 00:19
running rakudo-stress step
[Coke] coleman: "ping rba or coleman or andinus" - is that list still correct? 00:21
coleman where is that written
For issues related to accessing the trinity.raku.org server (where the release files are hosted), it's me and andinus 00:23
[Coke] it's in rakudo's docs/release_guide.pod 00:28
please wait until post release but please update that phrasing.
coleman will do 00:31
[Coke] (asking you in case you see anything else suss in the trinity section) 00:32
TODO passes in 6.c errata
(which I assume is fine given the nature of that branch) 00:33
jdv todos, yeahl 00:47
99% of the time actual breakage is just someone didn't port the test changes back
[Coke] on rakudo-post-test step 00:50
if I'm reading this right, don't run human-upload, but cd to the nqp & rakudo dirs to push commits/tags and manually upload tarballs to trinity 00:51
jdv i use human-upload 00:53
iirc it pushes the git changes up
the "upload" part doesn't work though and times out
that is the trinity scp parts 00:54
jdv hmm, formatting error, that second scp should be on hte next line 00:55
i think i have that fix in my local and forgot to push it
[Coke] ok. I know the scp will fail because I have no key for trinity on this box.
Geth nqp/main: 454c226b18 | (Will Coleda)++ | tools/templates/MOAR_REVISION
[release] Bump MoarVM revision to 2025.01
 00:56
nqp/main: 6ed2641d5a | (Will Coleda)++ | VERSION
[release] Bump VERSION to 2025.01
rakudo/release-2025.01: dec3f9516a | (Will Coleda)++ | tools/templates/NQP_REVISION
[release] Bump NQP revision to 2025.01
rakudo/release-2025.01: 5545d004a5 | (Will Coleda)++ | VERSION
[release] Bump VERSION to 2025.01
[Coke] jdv: it's two lines here? 00:57
ah, it's not indented so it's flowing
jdv yeah...
[Coke] wonders how long human-upload will hang trying to scp
[Coke] there we go. connection timed out on scp... doing those uploads manually... 00:59
not a zippy cxn to trinity 01:04
files uploaded to trinity 01:05
Geth rakudo: coke++ created pull request #5771:
Release 2025.01
rakudo/main: 4 commits pushed by (Will Coleda)++ 01:07
[Coke] jdv: what is the token in tools/releaseable/github-release.p6? 01:09
jdv moarvm.org/releases.html
maybe you didn't format it the same as usual?
oh, just a rando pers gh token 01:10
[Coke] jdv - used the markdown tool; Will clean that up.
jdv my token is just a "gh classic personal access token" with "repo" scope 01:12
there's a markdown tool?!
it's likely because the addition to this is different that the rest of it: github.com/MoarVM/MoarVM/blob/main.../ChangeLog 01:15
[Coke] "make sure you have text::markdown installed"...
ok.
when running github-release.p6 - same announcement file for both nqp & rakudo? 01:16
also - is "-asset" necessary or is that part of the manual fixup after?
jdv i think its required to make the dumb script work 01:17
and it times out again
[Coke] Server responded with 404 Not Found (POST api.github.com/repos/raku/nqp/releases) 01:18
jdv sure 01:19
[Coke] (and it didn't seem to do anything to github.com/raku/nqp/releases) 01:20
jdv how'd you call it? 01:21
here's how it called it last time - gist.github.com/jdv/fc753c72d9e23b...e193b2b82c 01:22
[Coke] sent you a privmst
hurm, no announce? ok.
jdv maybe that post is because your token isn't working? 01:24
rather 404 on post
[Coke] OK, got a 400 bad request after mucking a bit with the token... 01:28
can you tell me what permissions your token has so I don't have to add them all? 01:29
jdv i did:) i just have "repo" checked
[Coke] I have given it admin:gpg_key, admin:org, admin:org_hook, admin:public_key, admin:repo_hook, admin:ssh_signing_key, audit_log, delete:packages, gist, notifications, project, repo, user, workflow, write:discussion, write:packages 01:31
ok, going back to just repo
jdv: Can you run the github-release for nqp & moarvm and I'll attach the files to them? 01:32
sorry 01:33
for nqp & rakudo
jdv moarvm is done already, the ci auto thingee replaced it but that's what i ran...yeah
i think you did nqp already
[Coke] @#($*&?
jdv not rakudo yet
idk
[Coke] so one of these runs worked, but I keep seeing 400/404s? 01:34
freaky
added the files to the nqp release. 01:35
OH. I wonder if the 400 was the asset upload error, and that's why the error went from 404 to 400 after the first call
jdv sounds close enough 01:36
weird your .asc file is smaller than mine
maybe your key is smaller or something
[Coke] weird.
releases created, files attached 01:37
jdv just realized the rakudo tar is 6.66MB 01:39
[Coke] 2 releases running, at least. 01:41
I thought you were telling me it was BIG, not that it was a fun number.
how long does it take for rakudo.org to restart? 01:45
(I pushed the commit but don't see it on the news page)
(after doing the ssh restart step)
jdv uh, the ssh command is the restart 01:47
[Coke] I think we're done?
jdv like immediately
just the moarvm formatting is all i know
hmm, did you push the stuff to rakudo.org and restart? 01:49
[Coke] rakudo.org/post/announce-rakudo-release-2025.01
Yes, I missed a colon. 01:50
fixed.
I have a page of handwritten notes to update the guide and/or open bugs
which is a job for tomorrow. 01:51
jdv i had that page for years and just merged it into the guide recently:)
[Coke] the more people we get to do releases, the better the guide will be!
jdv github.com/MoarVM/MoarVM/blob/main.../ChangeLog and www.moarvm.org/releases.html are still not right
i'm not sure who will do next month's release yet. maybe patrickb or you or me... 01:52
[Coke] Right, I did the formatting based on cutting and pasting something, which screwed that up. Not worth re-releasing, but worth fixing. 01:53
I put Coke|jdv on the "next".
Geth rakudo: dontlaugh++ created pull request #5772:
Indicate how to ask for write access to the file server
[Coke] But the fact that I was able to do it (mostly clean) is a good sign
jdv yup
how different was the ones you did way back then? 01:54
coleman My PR to rakudo's docs: to request ssh access to Trinity (our file server), open an issue on Raku/infra
[Coke] the Ake stuff simplified a lot of it 01:57
Geth rakudo/main: e123587f33 | (Coleman McFarland)++ (committed using GitHub Web editor) | docs/release_guide.pod
Indicate how to ask for write access to the file server
rakudo/main: b891c808c7 | (Justin DeVuyst)++ (committed using GitHub Web editor) | docs/release_guide.pod
Merge pull request #5772 from dontlaugh/patch-1

Indicate how to ask for write access to the file server
jdv ah, cool
yeah, its pretty nice. sometimes installing ake goes bad cause deps or something 01:58
[Coke] ok, I think the moarvm.org releases.html page is fixed in the repo, but not on the site? 02:03
ah, 5-10 minutes
and that is totally on me, I did not check my local copy (I think I was on azure at that point) 02:04
jdv huh?
oh, its fixed inn your fork 02:05
can you push to the main repo yet or just create a pr?
[Coke] DOH
forgot that. :)
PR created 02:06
found "rakudo perl 6" on more than thing going through the releae.
jdv merged
sure:) 02:07
moarvm.org is fixed. yay. dinner. afk & 02:10
thanks!
[Coke] \o/ 02:14
Geth rakudo/main: 11 commits pushed by (Elizabeth Mattijsen)++
review: github.com/rakudo/rakudo/compare/b...35e95ffc9c		 10:59
roast: 9da1f354d9 | (Elizabeth Mattijsen)++ | 4 files
Adapt tests to handle BUILDALL -> POPULATE migration

These tests where hardcoded to check / ignore BUILDALL as a method name. Since Rakudo ea1831c79fc21ea3e582 the generated method to build an instance is called POPULATE, so checks need to check for that if a method named POPULATE can be called.		 11:03
rakudo/main: b1ce6561d9 | (Elizabeth Mattijsen)++ | 3 files
Make DateTime.clone do the right thing for subclasses/mixedins

This was at the root of #1238 for DateTime. Also mark &!formatter as built(:bind) so that we don't need to put a container in there to make POPULATE do the right thing.		 11:37
lizmat BTW, [Coke]++ jdv++ for the 2025.01 release! 11:38
Geth roast: dd3202f302 | (Elizabeth Mattijsen)++ | 4 files
Add tests for #1238
 11:53
¦ rakudo: lizmat unassigned from zoffixznet Issue IO::CatHandle.read switches handles too early github.com/rakudo/rakudo/issues/1313 12:03
[Coke] ok, everything quiet the day after the release is nice. ;) 16:44
lizmat [Coke]++ thanks again, sorry I wasn't around to be of help 17:04
Geth rakudo/main: bbd277079e | (Will Coleda)++ | docs/release_guide.pod
Updates to release guide from 2025.01 release
 17:52
[Coke] lizmat: No, it was fine, no worries.
the guides were pretty good, just had a few confusions (and will know to just do it on the linux box first next time) 17:53
Opening some tickets today and updating the release guides.
lizmat what box where you trying on before that? 17:54
spectest: Files=1359, Tests=119777, 211 wallclock secs (17.50 usr 3.55 sys + 1217.01 cusr 72.20 csys = 1310.26 CPU)
stresstest: Files=1415, Tests=198208, 254 wallclock secs (21.36 usr 4.03 sys + 1472.96 cusr 81.37 csys = 1579.72 CPU) 17:55
[Coke] my mac
worked great until "make the actual tarball"
lizmat ah, so we can't do a release from a Mac ?
[Coke] at which point, the perl script (why are we using perl for a critical part of our release infra?) barfed on locale settings, and getting around those, the tar command is using a LOT of what I'm guessing are gnu specific args 17:56
lizmat: correct. Looks fixable, though.
Just not in the pressure of doing a release.
lizmat ack... and why perl: I do not know, seems also fixable :-) 17:57
[Coke] updating the guides where it makes sense, opening tickets where needed. 17:58
Geth rakudo/main: e4e5655425 | (Justin DeVuyst)++ | docs/release_guide.pod
clarify path on build sanity check
 17:59
[Coke] jdv++ 18:00
jdv well, with limiited resources, priorities...:)
wrt the release on $rando_os
maybe i'll try to do the release from a obsd box next time;) 18:02
[Coke] github.com/MoarVM/MoarVM/issues/1902 18:06
for the perl issue - it's just on Moar. I think we're already avoiding this entirely on rakudo
Geth nqp/main: 6f7450e9a8 | (Elizabeth Mattijsen)++ | tools/templates/MOAR_REVISION
Bump MoarVM to use siphash instead of rapidhash

  MasterDuke++		 18:49
rakudo/main: e413a6e2d7 | (Elizabeth Mattijsen)++ | tools/templates/NQP_REVISION
Bump NQP to use rapidhash instead of siphash for string hashes

  MasterDuke++		 18:59
nqp/main: 087b5628ac | (Elizabeth Mattijsen)++ | tools/templates/MOAR_REVISION
Bump MoarVM to speedup find_cclass, MasterDuke++
 19:02
rakudo/main: 3b06479846 | (Elizabeth Mattijsen)++ | tools/templates/NQP_REVISION
Bump NQP to speedup find_cclass, MasterDuke++
 19:11
nqp/main: f506e924d6 | (Elizabeth Mattijsen)++ | tools/templates/MOAR_REVISION
Bump MoarVM to optimize CPointer passing to NativeCall

Timo++		 19:18
rakudo/main: 0cad3547bf | (Elizabeth Mattijsen)++ | tools/templates/NQP_REVISION
Bump NQP to optimize CPointer passing to NativeCall

Timo++		 19:26
nqp/main: 88b290864a | (Elizabeth Mattijsen)++ | tools/templates/MOAR_REVISION
Bump NQP to bump libuv to 1.50, MasterDuke++
 19:33
rakudo/main: 4350d5cacf | (Elizabeth Mattijsen)++ | tools/templates/NQP_REVISION
Bump MoarVM to bump libuv to 1.50, MasterDuke++
 19:44
nqp/main: fe1884a821 | (Elizabeth Mattijsen)++ | tools/templates/MOAR_REVISION
Bump MoarVM to counteract drift in timers with repeat

Timo++		 19:50
rakudo/main: c250768b28 | (Elizabeth Mattijsen)++ | tools/templates/NQP_REVISION
Bump NQP to counteract drift in timers with repeat

Timo++		 19:59
nqp/main: f873bf93b4 | (Elizabeth Mattijsen)++ | tools/templates/MOAR_REVISION
Bump MoarVM to enable experimental c11 atomics on MSVC

  ugexe++		 20:03
rakudo/main: 2be9b080e1 | (Elizabeth Mattijsen)++ | tools/templates/NQP_REVISION
Bump NQP to enable experimental c11 atomics on MSVC

  ugexe++		 20:08
nqp/main: cfe3119a02 | (Elizabeth Mattijsen)++ | tools/templates/MOAR_REVISION
Bump MoarVM for debugserver loaded file notifications

Timo++		 20:13
timo oh 20:17
lizmat: can you give me a shout before you merge the next of my pull requests so i can check if it should be a draft pr instead? 20:18
lizmat will do
I thought it was ready (enough) and had been sitting there for a while now 20:19
#1883 ??
also #1882 will appreciate some TLC from you :-) 20:20
timo well, the CI is red on that merge request
needs to be re-run because the logs have been purged
lizmat re-runs requested 20:21
Geth rakudo/main: af3b5a6524 | (Elizabeth Mattijsen)++ | tools/templates/NQP_REVISION
Bump NQP to get debugserver loaded file notifications

Timo++		 20:22
[Coke] Q: how do the individual NQP bumps help? 20:23
individual CI runs?
timo for bisecting
[Coke] +1 20:24
timo theoretically, we could build stuff into the bisect bots / scripts that goes into NQP versions when the issue happens to hit a bump commit 20:25
lizmat hmmm..looks like the REA harvester dies on HEAD now :-( 20:32
timo i don't think i saw the branch with the arm fix be merged? 20:34
[Coke] github.com/rakudo/rakudo/issues/5606 - this like a DIHWIDT - theyre trying to use utf8 on something that isn't necessarily
lizmat timo: which one is that ?
timo i think it doesn't have a pull request maybe 20:35
lizmat there's no PR for it?
nuking ~/.raku/precomp fixed the problem
timo there is now a PR 20:39
lizmat thanks, will merge later today or tomorrow, after CI is complete 20:40
timo the problem that the PR is fixing is nondeterministic
bumping revisions makes the CI very dumb 20:42
lizmat ok, so you're saying I should ignore CI in this case ? 20:43
or that I shouldn't be bumping so much ?
timo neither
we need to do something that makes the CI not turn beet red whenever there is multiple bumps in a row
we might be able to grab the exact commits for the three repos at the very start of pipelining, even before any of the builds start? like, in a separate stage even 20:44
lizmat that would be nice :-)
[Coke] huh. if I search for issues like >> is:issue state:open -label:jvm -label:"Will be addressed in RakuAST" -label:"Fixed in RakuAST" << 21:05
I still get the will be addressed labels in the search
Geth rakudo/main: 997ca6a464 | (Elizabeth Mattijsen)++ | src/core.c/Array.rakumod
Make sure Array.clone remembers deleted slots

Attempted to be smart about it initially, but these all failed because you can not reliably clone an iterator and its internal state correctly. So in the end, just check for nulls in the original reified, and copy those.
Also make the code a bit more readable.
Fixes R#1434		 21:34
linkable6 R#1434 [open]: github.com/rakudo/rakudo/issues/1434 [data types] Array.clone does not clone holes correctly
Geth roast: 2221ae07d6 | (Elizabeth Mattijsen)++ | S02-types/array.t
Add test for R#1434
 21:39
linkable6 R#1434 [open]: github.com/rakudo/rakudo/issues/1434 [data types] Array.clone does not clone holes correctly
lizmat afk& 21:40
[Coke] where would [max] be in core? 23:06
timo that should just be METAOP_REDUCE applied to &infix:<max> 23:09
[Coke] m: max [^100000000] 23:14
camelia WARNINGS for <tmp>:
MoarVM panic: Memory allocation failed; could not allocate 235446976 bytes
[Coke] wondering if we can make that old ticket work, but we'd probably have to special case on ranges *and* the function. 23:15
timo well when you shove ^10000000 into an array by using the [] constructor syntax, you are actually asking it to create an array with one entry for all these numbers in that range 23:18
[Coke] right. I'm wondering if there's a way to skip the array creation there in special cases; guessing this is just something we should document as a trap (it already kind of is) and close out the ticket. 23:19
github.com/rakudo/rakudo/issues/1210
timo not sure if it'd be possible to special-case-handle `max [^bignum]` except when it's very literally that 23:35
i imagine, most cases the array would have to be created and filled long before we even realize it was only being fed into max or min 23:36
and if we make that literal case work and someone refactors that code ever so slightly, the optimization breaks and suddenly your program will consume huge amounts of memory 23:37
[Coke] so let's just close that ticket out. 23:43
timo well, the ticket is about more than just that? 23:56

Powered by Cro and the Raku® Programming Language.

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

Thank you!