🦋 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. |
|||
00:27
SmokeMachine left,
SmokeMachine joined
09:37
finanalyst joined
09:44
sena_kun joined
|
|||
Geth | nqp/main: 45dfc92fa3 | (Elizabeth Mattijsen)++ | tools/templates/MOAR_REVISION Bump MoarVM to get timo++ patrickb++ fixes - optimizing nativecall - fixing long path issue on Windows |
09:57 | |
rakudo/main: 1fe089e36d | (Elizabeth Mattijsen)++ | tools/templates/NQP_REVISION Bump NQP to get latest MoarVM fixes |
10:07 | ||
rakudo/lizmat-post2025.01: 0429be00ab | (Elizabeth Mattijsen)++ | 5 files Lose Positional argument to BUILDALL This makes Date instantiation (other than Date.today) about 45% faster, because it uses the direct POPULATE interface. And drops the spectest time from 211 -> 208 wallclock (1308 CPU -> 1279 CPU seconds). The standard object instantion uses the following call sequence: ... (23 more lines) |
11:17 | ||
rakudo/lizmat-post2025.01: 351c4cca7e | (Elizabeth Mattijsen)++ | src/core.c/Supplier.rakumod Simplify Supplier creation Just specifying the attribute correctly is now enough to get a $!taplist attribute set up correctly without needing to resort to bless/BUILD shenanigans |
11:42 | ||
rakudo/lizmat-post2025.01: 747a50fcbb | (Elizabeth Mattijsen)++ | src/core.c/Backtrace.rakumod Use POPULATE interface for Backtrace::Frame Just because we can |
12:08 | ||
rakudo/lizmat-post2025.01: 294e485a8b | (Elizabeth Mattijsen)++ | lib/MoarVM/Profiler.rakumod Use POPULATE interface on MoarVM::Profiler Because we can |
12:16 | ||
lizmat | m: sub a(int() $a) { dd $a }; a 42 | 12:43 | |
camelia | Impossible coercion from 'Int' into 'int': method int returned an instance of BOOTInt in sub a at <tmp> line 1 in block <unit> at <tmp> line 1 |
||
lizmat | meh | ||
13:35
kjp left
13:59
finanalyst left
15:02
guifa joined
15:11
sena_kun left,
sena_kun joined
15:15
sena_kun left
15:16
sena_kun joined
15:47
guifa left,
guifa joined
16:09
finanalyst joined
16:55
guifa left
17:32
guifa joined
17:57
guifa left
18:04
guifa joined
18:25
guifa left
|
|||
[Coke] | late start today (lots of stuff going on in the house), starting release work in about 15m | 18:26 | |
going to rollback that last minute merge to MoarVM/main | 18:27 | ||
jdv | 2 bumps too:( | 18:39 | |
[Coke] | eh. was going to have to update the bumps anyway | 18:40 | |
jdv | really? | ||
[Coke] | Those can just move forward, I think | ||
I didn't see them, but surely I have to make the bumps to the tags, and I doubt the changes just now did that? | 18:41 | ||
jdv | i mean, aside from release bumps. i guess you can bump "backwards" | ||
the release tool does the release bumpings | |||
[Coke] | right, but why bump back and then bump to tag? | ||
jdv | will spectest work with the bumps and the rolled back moarvm | 18:42 | |
that is the question | |||
[Coke] | I might be missing a step here. I'll ask if I get stuck going through the directions | ||
jdv: yup, that's a good question | |||
eh, not a big deal | |||
jdv | sure. but won't you be spectesting the wrong build at some point? | 18:43 | |
anyway, good luck! | 18:44 | ||
[Coke] | ... it's not a big deasl *to fix the bumps* | 18:50 | |
so I'll do that and then make sure I'm testing the right thing. | |||
... Looks like a BUNCH of merges on thursday also. | 18:51 | ||
ok, 2. and those both look very safe. no worries there. | 18:52 | ||
NQP is Copyright (C) 2009-2020 by The Perl Foundation. See F<LICENSE> | 19:05 | ||
for licensing details. | |||
Looks like haven't done that one in a while. Updating it to TRF, 2025, complain soon if that's a problem. | 19:06 | ||
Geth | nqp/main: c64ad92e67 | (Will Coleda)++ | README.pod Update copyright notice, use other foundation aka |
19:07 | |
rakudo/main: ae02996e21 | (Will Coleda)++ | README.md update copyright year, use foundation alias |
19:09 | ||
[Coke] | jdv: Update Rakudo’s leap-second tables: --- that script is gone | 19:12 | |
19:13
finanalyst left
|
|||
[Coke] | fixing. | 19:13 | |
jdv | good thing there's no leap second:) | 19:14 | |
Geth | rakudo/main: 0eea1f3b19 | (Will Coleda)++ | docs/release_guide.pod Update leap second release info |
||
[Coke] | .. yes, because it's not as simple as just running 'add-leap-second' anymore. :( | 19:15 | |
Looks like the logic for checking the web site didn't make it into the raku version | 19:16 | ||
Geth | ¦ rakudo: coke self-assigned tools/add-leap-second.raku is useless github.com/rakudo/rakudo/issues/5769 | 19:20 | |
rakudo/main: 7f2a62acb1 | (Will Coleda)++ | docs/release_guide.pod Update notes about leap second which has been a manual operation for some time |
19:28 | ||
[Coke] | jdv - do you use 'z' where it's suggested in the release guide or do you do those steps manually? | 19:42 | |
jdv | i do not use "z" | 19:50 | |
its suggested only for blin prep afaik | 19:53 | ||
19:55
finanalyst joined
|
|||
[Coke] | ok. looks like it might be nifty for the bumps (except I'm getting a SEGV at the moment. :( | 20:07 | |
Geth | nqp/main: 42889b1fc6 | (Will Coleda)++ | tools/templates/MOAR_REVISION Bump MOAR |
20:10 | |
rakudo/main: db4deb1651 | (Will Coleda)++ | tools/templates/NQP_REVISION bump NQP |
20:11 | ||
[Coke] | jdv: naming convention for "checkout a new release branch." ? | 20:12 | |
jdv | whatever is the last one? | 20:14 | |
[Coke] | found it | ||
jdv | that usually holds for most stuff | ||
Geth | rakudo/main: 81ddecc11a | (Will Coleda)++ | docs/release_guide.pod note naming scheme for release branch |
20:15 | |
[Coke] | jdv: thoughts on February? | 20:25 | |
(just found a sneaky "rebuild and retest everyting" in the moarvm readme..) | 20:26 | ||
jdv | oh | ||
how is it sneaking up on you? | |||
[Coke] | t/spec/S32-io/IO-Socket-Async.t (Wstat: 256 Tests: 3 Failed: 1) Failed test: 2 | 20:27 | |
Wasn't expecting to have to do multiple builds with various moarvm configs. Missed it on previous readthroughts. | |||
jdv | i guess 2/22 is fine | 20:28 | |
i don't do multiple builds | |||
[Coke] | ./install/bin/raku t/spec/S32-io/IO-Socket-Async.t | ||
Unhandled exception: Missing or wrong version of dependency 'gen/moar/stage2/NQPHLL.nqp' (from 'gen/moar/Pod.nqp') | |||
moarvm, release guide, making a release, #3 | |||
jdv | i do that just with the main rakudo build parts lying around for the dsay | 20:29 | |
idk why you'd need to rebuild | |||
[Coke] | ? | ||
the part where it says "after recompiling..." | 20:30 | ||
jdv | ah, that says "if possible" so its never been possible for me:) | ||
[Coke] | make t/spec/S32-io/IO-Socket-Async.t is repeatable error on test #2 here. | 20:31 | |
jdv | that sucks. when did it last pass? | ||
i'll try locally. my box is slow. | 20:35 | ||
Geth | rakudo/release-2025.01: 1121aefc74 | (Will Coleda)++ | docs/release_guide.pod Add this/next release, update copyright |
20:36 | |
[Coke] | Might just be that I'm doing this on a mac. | ||
timo | [Coke]: can you set RAKUDO_MODULE_DEBUG=1 to see where it's looking for the Pod.nqp precompiled file that's wrong? | ||
[Coke] | if I get a clean run from someone on linux, fair play | ||
timo | you may have to nuke the prefix before building stuff | ||
[Coke] | timo: gist.github.com/coke/2057ad1132c8a...70399a928e | 20:38 | |
timo: so do a rm -rf ./install and try the whole thing again? | |||
timo | does your ./install/bin/raku work at all? | ||
[Coke] | That's a lot to track down this thing I can avoid by running 'make <spectest>' | 20:39 | |
timo | is that actually really your prefix? and you already did "make install"? | ||
[Coke] | I didn't set a prefix, using the default | ||
... and no, ./install/bin/raku -e 'say 3' also barfs. | |||
Lovely | |||
timo | "make <spectest>" uses the not-yet-installed raku | ||
[Coke] | rebuilding... | ||
timo | so you can run the spectest before installing | ||
stop | |||
jdv | sounds like a personal problem:) | ||
[Coke] | whew. | 20:40 | |
I was right, that *was* a familiar error. :) | |||
timo++ | |||
Unhandled exception in code scheduled on thread 4 | 20:41 | ||
address already in use (address: 0.0.0.0:5000) | |||
timo | might still be running another instance of that test, or some random program uses port 5000 for itsefl | ||
[Coke] | Guessing this is an OSX spectest issue - jdv, let me know if you have similar issues. | ||
ah, yes ,works great with another port. | |||
I picked '6583' at random, test passed. | 20:42 | ||
timo | it should really use a randomized port. i think if you pass 0 as the port it's supposed to give you a random free port | ||
jdv | that particular spectest works for me after just building head and full spectest running... | ||
[Coke] | timo: if I change it to a literal 0, it dies in the same spot here on mac. | 20:47 | |
timo | with "already in use"? you would also have to change the code to ask for what exact port it did end up getting | 20:48 | |
[Coke] | can we close github.com/Raku/roast/issues/320 ? | ||
timo: fails in *same place*, sorry | |||
jdv | full spectest pass here | ||
[Coke] | or at least, very early. | 20:49 | |
timo: I opened github.com/Raku/roast/issues/868, fwiw. | 20:50 | ||
timo | good | ||
[Coke] | tools/create-release-announcement.raku | 20:52 | |
Unable to call postcircumfix @lines[ (Any) ] with a type object | |||
jdv: do you use the script to create the release announcement or do you do it by hand? | |||
jdv | i use hte script but i noted there's a bug or more | ||
[Coke] | ... if you update the Planned future releases too soon, it breaks the script. | 20:56 | |
jdv | i've never deviated too far off "script" so i wouldn't know. sorry? | 20:58 | |
21:04
kjp joined
|
|||
Geth | ¦ rakudo: coke self-assigned create-release-announcement error github.com/rakudo/rakudo/issues/5770 | 21:11 | |
[Coke] | jdv: ... you also may be assuming some things about the script. :) | 21:14 | |
there's no step that says "add the changlog data to the release announcement - but I think that's implied by the "script is broken" step, yes? | 21:15 | ||
Also: are you using 'ake' ? | |||
(guessing so as those instructions are pretty specific) | 21:17 | ||
jdv | yes, i use ake | ||
[Coke] | ah, found the instruction I thought wasn't there about wiki, nevermind. | 21:18 | |
jdv | that's in the end of...yeah | ||
there's plenty to fix and/or automate if anyone is interested:) | |||
[Coke] | Yup, I have several tickets I'll work on in the next month or so as a result. | 21:19 | |
jdv: does "Create the release commit" just mean "all the changes so far"? I have actually been doing intermediate commits once I did the branch. | 21:21 | ||
Also: do we need the "deliberately not logged" ? | |||
jdv | all those changes from the release instructions typeically end up in one commit | 21:22 | |
if you look at the last one you'll see | |||
a006a3d1d7c633f216ffd24b5412e9b88b70554a looks like it | |||
linkable6 | (2024-12-14) github.com/rakudo/rakudo/commit/a006a3d1d7 Update changelog + announcement | ||
[Coke] | ok, but that's not a requirement, I'm guessing, more stylistic? (I've already pushed a small one, can force commit to fix it if needed) | 21:23 | |
jdv | where is this deliberately not logged stuff? | ||
[Coke] | More concerned about the "deliberately not logged" | ||
in the commit message | |||
jdv | oh, i think that's for releasable? | 21:24 | |
[Coke] | ah, needed by the release bot | ||
(it's mentioned MUCH earlier in the release guide) | |||
jdv | i think the assumption is that one would follow the guide unless one knew better or had a very good reason to deviate and knew what they were doing? idk. | 21:25 | |
that's more or less how i've been doing it | |||
Geth | rakudo/release-2025.01: 0719ab0728 | (Will Coleda)++ | 3 files Update changelog + announcement Deliberately not logged: [cacc8761][aeca75cf][3efc376a][f58bed9f][49d06524][d87aa6f5] [cfa01ab4][79fb513a][f12ff834][0bf65ec7][6363b8e1][c3d9b3d5] [dc02fbf4][633df289][b4d79832][1bd183d4] |
21:27 | |
[Coke] | Yup, that is in there, just needs a slight edit so that it's referenced also at the point where you need it | 21:28 | |
jdv: why are the moarvm release branches "02024.010", e.g? (why the extra 0?) | 21:43 | ||
er, 2024.010 | |||
(the extra zero is only at the end) | |||
jdv | i'd be guessing | 21:49 | |
[Coke] | Following the convention (will update that release guide to mention it, unless we agree to drop it going forward) | 21:50 | |
perl: warning: Setting locale failed. | 21:53 | ||
perl: warning: Please check that your locale settings: | |||
(trying a make release on moarvm) - digging | |||
jdv | looks like it was introduced by my predecessor: github.com/MoarVM/MoarVM/commit/b7...2aa6923R58 | 21:54 | |
the release guide does mention it? isn't that why you're doing it:) | 21:55 | ||
your box seems crankier than mine | |||
21:56
guifa joined
|
|||
[Coke] | the release guide contradicts the list of release branches | 21:58 | |
I may have to cut the actual release on a linux box. :| | 21:59 | ||
will try on the ubuntu machine I have access to | |||
... crap. that won't work, as I can only do signed commits from here. | 22:00 | ||
oh, it's hardcoded into the release.sh file? | 22:02 | ||
jdv | you don't have utf8? | 22:04 | |
22:05
kjp left
|
|||
[Coke] | I'm clearly using utf8. that particular setting causes perl to barf. | 22:05 | |
22:05
kjp joined
|
|||
[Coke] | if I change it locally to LC_ALL=C, the script runs, but then dies again: | 22:05 | |
./tools/release.sh 2025.01 | |||
fatal: ambiguous argument '2025.01': unknown revision or path not in the working tree. | |||
if I use 2025.010 (matching the branch name), I get a bunch of errors, making a gist | 22:06 | ||
gist.github.com/coke/da5b5e8c3c22c...0ba192c31d | 22:07 | ||
jdv | i have never run into that before | 22:08 | |
[Coke] | guessing it's using a bunch of tar options that aren't available here. | ||
so, step 0: must cut releases from linux? | 22:09 | ||
jdv | i know it all works on debian. that's all i got. | 22:10 | |
[Coke] | ok, I'm stepping away for 15m so I don't freak out, brb. | 22:13 | |
timo | what the heck :) | 22:25 | |
missing features from the make implementation? using sh instead of bash? | 22:26 | ||
or a bsdtar instead of gnu tar or something | |||
it does look like it's interpreting all these arguments as filenames instead of flags | 22:27 | ||
[Coke] | It's still better than the original christmas release. :) | 22:29 | |
I'm creating the keys I need on the azure linux box now | |||
jdv | m$ hosted linux, i would have never thought of it decades ago | 22:30 | |
hopefully you can just copy the dirs over | |||
[Coke] | gpg: signing failed: Inappropriate ioctl for device | ||
this in a forked repo I have just to test "git tag -as TESTTAG" | 22:31 | ||
jdv | gotta whip those boxes into shape:) | 22:32 | |
[Coke] | github.com/coke/json_fast/tags - verified, whee | 22:33 | |
tar: MoarVM-2025.01/tools/update_ops.raku: implausibly old time stamp -9223372036854775808 | 22:35 | ||
... the tar xvf still worked, though | |||
make: src/core/args.c: Timestamp out of range; substituting 2514-05-30 01:53:03.999999999 | |||
timo | what on earth are you doing :D | 22:36 | |
[Coke] | (doing step 9 of the moarvm release - everything is still moving...) | ||
FOLLOWING THE RELEASE GUIDE. ;) | |||
timo | how novel | ||
[Coke] | make: warning: Clock skew detected. Your build may be incomplete. | 22:37 | |
drwxrwxr-x 3 azureuser azureuser 4096 Jan 25 22:34 docs | |||
just a random line from ls -l - those dates seem fine.. | |||
my date is right... | 22:38 | ||
timo | ah, we grabbed the files from azure build? | ||
of course there is more than just one date for a file | |||
`stat -c '%w %x %y %z' filenames` | 22:39 | ||
[Coke] | ? I'm on my own azure box | 22:40 | |
This doesn't have anything to do with the CI buidls | |||
github.com/MoarVM/MoarVM/tags - anyone want to verify anything about that before I keep moving? | 22:41 | ||
jdv | maybe clock or fs diffs? you copied them over? | 22:43 | |
22:44
sena_kun left
|
|||
jdv | i can't build it:( | 22:44 | |
[Coke] | like, at all? | 22:46 | |
from the tag? | |||
jdv | yeah. weird. | 22:47 | |
can you untar your tar and it works? | |||
[Coke] | I haven't pushed the tar file anywhere yet | ||
the ones up there are autogenerated. | 22:48 | ||
trying the tar.gz .. | |||
timo | that is now 2025.10 rather than 2025.010? | ||
[Coke] | make: *** No rule to make target `3rdparty/cmp/cmp.h', needed by `src/main.o'. Stop. | ||
01 | |||
the *branches* had the extra zero for some reason | |||
jdv | i can't build last month's either so not i'm a bit confused | 22:49 | |
[Coke] | ok, why does that tag not have the 3rd party stuff? | ||
jdv | yeah, that's what release.sh does i think | ||
[Coke] | nobody is using those, are they? | ||
timo | yeah github doesn't put submodule contents in these "sources" tarballs | 22:50 | |
jdv | i'd say its ignorable as long as the tarball you gen works | ||
[Coke] | I tried 2024.05 at random - also missing 3rdparty | ||
jdv | yeah, i forgot about that | ||
[Coke] | ok, let me put that somewhere other than the official site. | ||
Can someone please grant me access to push to moarvm.org? | 22:54 | ||
(i'll put them in the right place but not create the announcement yet) | 22:55 | ||
jdv | maybe hte infra guys? | ||
it used to be jnthn or liz maybe | |||
[Coke] | coleman, can you grant me write access to moarvm.org? | 22:56 | |
jdv: do you see "admin" or "settings" on the github repo? | |||
coleman | in the server? | ||
jdv | on what repo? | ||
coleman | I don't have GitHub access to moarvm repo | ||
[Coke] | moarvm.org | ||
but probably not that either. :| | 22:57 | ||
jdv | oh, right. its just gh. you can tell i'm awake. | ||
i do not | |||
[Coke] | ok. Anyone have a place I can upload these files so maybe jdv can put them in the repo? | ||
... I'll throw it in a github repo of my own. one sec | 22:58 | ||
jdv | are you wanting to push to github.com/MoarVM/moarvm.org? or upload the 2 files to github.com/MoarVM/MoarVM/releases/tag/2025.01? | 22:59 | |
[Coke] | github.com/coke/moarvm.org/tree/master/releases | 23:00 | |
I was trying to upload them to moarvm.org - which I just forked. | 23:01 | ||
so the 2025.01 moarvm tar and asc file are in my fork right now | |||
If we're happy, I can do the work in my fork for the website and create a PR | |||
(which jdv can then merge) | |||
(waiting for ake to explode when I get there. :) | 23:02 | ||
If that tar file is bad for y'all, we may have to delay a day | 23:03 | ||
jdv | gpg: Can't check signature: No public key | 23:06 | |
:) | |||
my tar spits out a bunch of implausibly old time stamp -9223372036854775808 | 23:07 | ||
and more datetime noise but ultimately "This is MoarVM version 2025.01 built with JIT support" | 23:09 | ||
[Coke] | is the gpg check just because I haven't shared it with you? | 23:10 | |
jdv | i think its cause your key isn't on a keyserver | ||
but more or less | |||
[Coke] | OK. github trusts it, at least. :| | ||
ok. so, are we OK with that level of datetime wonkiness? no doubt it will recur on the others. | 23:11 | ||
jdv | we can try. who knows if downstream will have real trouble. | ||
[Coke] | ok. finishing up the moarvm.org stuff. | 23:12 | |
jdv | you can't just untar and retar to fix it? idk. | ||
[Coke] | looks like the assets on the release include a lot of stuff by default that wasn't in the last release - did you remove, e.g. MoarVM.macos-13.tar.gz | 23:17 | |
huh - no, I see them in 2024.10 | |||
jdv | last release the ci automation thing broke so i did it by hand, old school style | ||
i think timo fixed it shortly thereafter | 23:18 | ||
or someone did | |||
you just have to add yours | |||
[Coke] | What about checking the "latest release" box when uploading the assets? yes? | 23:20 | |
jdv | got me | ||
i just upload the 2 files | 23:21 | ||
coleman | it's good to click that for Official releases | ||
it's just metadata to distinguish from pre-releases | |||
jdv | we don't do anything but official releases | 23:22 | |
[Coke] | so check it every time? :) | ||
will add it to the list | |||
coleman | yea, but that's what the box means | ||
[Coke] | MoarVM release ... done, barring merging my PR on moarvm.org | ||
.. .and the PR to merge the release branch back to main | 23:24 | ||
jdv | i have to run for a while. this is running exceptionally later than i had thought it would:( | 23:25 | |
[Coke] | Yup, sorry, I got a late start! :( | 23:27 | |
want me to not press the button on rakudo until you return? | |||
coleman - upcoming step requires me to scp to trinity | 23:37 | ||
coleman | i believe your pub key was uploaded last month | 23:38 | |
[Coke] | should I be able to ssh in or just scp? | ||
coleman | rsync or scp | 23:39 | |
[Coke] | ... I'm going to have bounce these files to my laptop from azure first, but that should work. | ||
coleman | github.com/Raku/infra/issues/59 | 23:40 | |
That documents what I've put on the server | |||
I think you can ssh as the rakudo.org user | 23:41 | ||
[Coke] | ah, really did need to specific an explicit key | 23:43 | |
coleman | If you need me to upload an additional key, it's no trouble. I'd just prefer not to | ||
But only to have One Less Key in the wild | 23:44 | ||
jdv | showered. now have to get groceries. maybe an hour. you're the rel mgr. if it looks ok, go for it! | ||
[Coke] | no worries, I'm good | 23:45 | |
woot | |||
coleman | cool; I'll be around all night. @ me if needed | ||
[Coke] | woot | ||
jdv | later | 23:46 | |
[Coke] | ... I should have ordered delivery to hit about now. | ||
coleman | ciao | ||
[Coke] | nqp release tagged (will take a minute for it to show up as a release) | 23:47 | |
OHAI we need gpg2? | 23:48 | ||
(easy fix) | 23:49 | ||
timo | i think so? | ||
[Coke] | I didn't need it for any of the github stuff | ||
but Ake is calling it specifically instead of gpg | 23:50 | ||
ok, working on rakudo build/test before pushing the release. | 23:51 | ||
Thanks for the support, folks! |