🦋 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. |
|||
jdv | who uses relocatable builds? i guess i don't because i could install zef the whole time. | 01:07 | |
japhb | [Coke]: Instant values are in TAI seconds, stored internally as an integer number of nanoseconds and then divided by 10^9 when looked up as something expecting seconds as a base. | 01:12 | |
jdv | this is now the second time in about as many months we've had a post release issue. not a great trend. | 01:15 | |
japhb | (It doesn't make a lot of sense for something of at least laptop size to have a concept of "instant" more precise than nanoseconds, because the average laptop is around a light-nanosecond across.) | ||
jdv: It sounded like this time we had a structural problem: we can produce builds configured in a way that our CI doesn't currently test. | 01:16 | ||
(I don't remember what last month's problem was off the top of my head.) | |||
jdv | it was the mimalloc on alpine thing. also could have been could by a apline ci build i guess | 01:17 | |
oddly enough both were caught during downstream release builds | 01:19 | ||
japhb | Well, downstreams are likely to have good CI for their own build configs. :-) | 01:31 | |
But yeah, it sounds like we need to expand our CI pool. | |||
At the very least it would be good to CI against any downstream that we want to consider "supported" or that we at least have a good relationship with (project to project) | 01:32 | ||
Doesn't mean we need to CI against everything on every commit, but it would be nice to have a "release CI pool" that is more all-inclusive | 01:33 | ||
s/all-// | |||
[Coke] | patrickb: maybe let's review the release docs and see how we can test a relocatable build in there. | 01:36 | |
Also we might want to consider starting doing a pre-release build. | 01:37 | ||
jdv | no, the manual release process is already too long | 02:14 | |
japhb | [Coke]: Is there a standard way to get such a pre-release to trigger CI from the downstreams? In other words, is this a part of standard project - distro handshake protocol? | ||
jdv | should we also try to build the release manually on alpine | 02:15 | |
? | |||
how many other things should the release manager be burdened with? | |||
both of the recent "downstreams" are "us" El_Che found the mimalloc one and patrickb found this one. | 02:16 | ||
japhb | Ah, I thought you meant distro build farms. | ||
jdv | it was just binary builds of some sort | ||
i think we've only had maybe one distro issue in a few years so that seems fine | 02:17 | ||
japhb | I guess then [Coke]'s idea comes into play: splitting out pre-release and release enough that El_Che/patrickb/whomever building/testing binary builds can be fit in between. | 02:19 | |
So it's not a huge added burden on the release manager, but it does split the release into possibly a two-day process. | |||
(Which of course may be its own burden) | 02:20 | ||
jdv | it would be extra work both up front and ongoing | ||
if we could reduce the existing burden more then maybe it could come out even in the end, maybe | 02:21 | ||
japhb | I certainly don't want to laser the RM, but we have to compare the effort of that versus the effort of doing additional point releases. | ||
nodnod | |||
jdv | i'd still like to know what is the impact here. with the last issue it was just alpine and i think El_Che just patched it up and we didn't need to re-engage the whole release chain | 02:24 | |
japhb realizes that he just ... yet AGAIN ... made the "increased baseline cost versus reduced risk" argument | 02:25 | ||
jdv | does this need to be fixed in 4 days or is in 4 weeks sufficient? | ||
yup | 02:26 | ||
japhb | I guess it depends on whether we consider "one of the downstreams will have to skip a month" to be a critical fail. | 02:28 | |
It's certainly not as bad as "we found a security issue", but it does feel like a broken window. | |||
patrickb | The e | 04:12 | |
The issue affects everyone using `rakubrew download` or any of the binaries released on rakudo.org. | 04:14 | ||
ugexe | which includes GitHub actions I presume | 04:30 | |
patrickb | I know too little about those, but possibly yes. | 06:19 | |
All in all, we're not breaking those people's workflow, but they just can't use 2025.06. | 06:20 | ||
Geth | rakudo/main: 74d02b5a30 | (Patrick Böker)++ (committed using GitHub Web editor) | src/core.c/Distro.rakumod Fix `raku -V` on strange non-standard MacOS (#5922) There exist non-standard MacOS versions out there (specifically MacOS-15 on Azure CI) that have this in their license file: SOFTWARE LICENSE AGREEMENT FOR APPLE CONFIDENTIAL macOS PRE-RELEASE SEED SOFTWARE ... (6 more lines) |
07:14 | |
lizmat | [Coke]: must have been asleep: the problem is that you'd need to also specify RAKUDO_RAKUAST=1 to make "use L10N:JA" work | 07:16 | |
I'll adapt the docs accordingly | 07:17 | ||
Geth | CY/main: b344bfe2f5 | (Elizabeth Mattijsen)++ | 4 files 0.0.7 |
07:30 | |
¦ NL: lizmat self-assigned Non-localized version information github.com/Raku-L10N/NL/issues/1 | 08:07 | ||
08:13
Geth left,
Geth joined
08:38
nine left,
nine joined
|
|||
[Tux] | `Type check failed for return value; expected Array[CSV::Field] but got Array ([CSV::Field.new(is_q...). You` | 09:56 | |
How can I prevent that `.new` is included in the type of the object being push on an array? | 09:57 | ||
lizmat | by not calling .new on it ? | 09:58 | |
[Tux] | is .new implicit? | 09:59 | |
lizmat | no | ||
[Tux] | so, then I am confused | ||
my $f = CSV::Field.new; @a.push: $f; | 10:00 | ||
lizmat | how is @a defined ? | ||
[Tux] | Array[CSV::Field] | ||
(it should, but /me would have to check all occurances) | 10:01 | ||
lizmat | my Array[CSV::Field] @a ?? | ||
[Tux] | my CSV::Field @a; | 10:02 | |
lizmat | ah, ok | ||
if you specify that as an argument, it expects an array that is typed that way, *not* an array that happens to have those types of values | |||
perhaps better: sub foo(@a where @a.are(CSV:Field)) ?? | 10:03 | ||
[Tux] | you lost me there | ||
lizmat | ah, return value | ||
[Tux] | to be clear on what I am trying to do now: I want to make field order stable for hashes | 10:04 | |
lizmat | the array that you return, how is that defined ? | ||
[Tux] | that means I have to remember the order of the first row and reorder the fields in method $f.fields | 10:05 | |
gist.github.com/Tux/dc845ac1f8fdae...611e76220d | 10:06 | ||
plus many more small changes, but specifically this part causes trouble | |||
Full current diff added to that gist | 10:07 | ||
lizmat | github.com/Raku/problem-solving/issues/482 | 10:36 | |
patrickb | I'd like to merge | 11:35 | |
github.com/MoarVM/MoarVM/pull/1812 | |||
That's the return prioritization PR from one and a half years ago. It's rebased to latest main, it passes all tests and spectest, including added tests to validate the new behavior. | 11:36 | ||
Will this cause any trouble because we haven't concluded how to deal with the 2025.06 fallout yet? | 11:37 | ||
lizmat | I would wait until that is cleared, and I would be in favour of a point release | 11:40 | |
patrickb | Ok. | 12:28 | |
lizmat | jdv ^^ | ||
jdv | are we sure the issue is fixed and no new issued have been introduced? i didn't notice any additional tests. | 13:26 | |
lizmat | irclogs.raku.org/raku-dev/2025-06-29.html#21:55 | 13:28 | |
5922 got also merged after some tweaks | |||
jdv | if we have a quorum then i can do a point release later today | 13:29 | |
lizmat | I guess +1 from lizmat patrickb | 13:30 | |
I guess [Coke] is on the road today | |||
jdv | [Coke]: ? | ||
lizmat | perhaps nine would like to chime in | ||
[Coke] as co-release manager ? | |||
jdv | oh, i didn't mean specifically rsc. just core-ish peeps:) | 13:31 | |
but, basically, yeah | |||
lizmat pings japhb ugexe | 13:34 | ||
jdv | looks like i didn't have enough coffee yet | 13:37 | |
lizmat | how so? | 13:38 | |
jdv | it looks like a point release later today unless someone gives it a thumbs down | ||
i missed your line on coke being on the road:) | 13:39 | ||
patrickb | I'll put some brains into which additional testing would be good in our CI. As about every one noticed - there is room for improvement. | ||
jdv | i have to run for a few hours. please don't commit anything crazy until after the release;) | 13:40 | |
patrickb | jdv: Will not. Thank you! | ||
jdv | patrickb: thanks | ||
lizmat: thanks | |||
patrickb | My current thinking is to not add more tests to the test suite, but exercise the installation with a few representative everyday situations and assert on those. E.g. install a module, make an actual web request via one of our web stacks, ... | 13:42 | |
lizmat | fwiw, most of the dists that I maintain, also try to install the dist in question as one of the CI tests | 13:43 | |
13:46
librasteve_ joined
14:02
vrurg_ joined
14:04
vrurg left
|
|||
patrickb | lizmat: I'm mostly aiming to exercise our integration, not the module installation itself. I.e. does an installed rakudo work, do the installed scripts work, does a standalone script using some module work, ... | 14:21 | |
lizmat | ah, ok :-) | 14:22 | |
patrickb | Integrations are pretty tricky to test, it's pretty much the stuff that usual tests never exercise, but bite you the instant you want to actually use the thing. | ||
lizmat | as we've found out :-( | 14:23 | |
patrickb | Which - in my defense - explains why I have a higher than average tendency to break releases. I work a lot on integrations. | 14:24 | |
[Coke] belatedly waves from the road. | 14:25 | ||
I vote "present" on a point release. I think in an ideal world we do it, but I also know I will not be able to help with it at all. | 14:26 | ||
16:35
librasteve_ left
|
|||
jdv | so it'll be 2025.06.01? | 17:03 | |
[Coke]: thanks | |||
Please halt all commits for a bit while we do an unscheduled point release which will commence shortly. | 17:04 | ||
lizmat is halting :-) | 17:05 | ||
jdv | i meant hold. a bit too dramatic there:) | 17:06 | |
lizmat | .oO( sorry, can't respond, I've halted :-) |
||
patrickb | 2025.06.1, (without the 0, right?) | 17:10 | |
lizmat | that indeed appears to be customary | 17:11 | |
although I do recall there being some issue with that | |||
? | |||
could very well be misremembering | |||
jdv | i don't think i care. 01 or 1? | 17:14 | |
i see more .1 than .01 so .1, sure. | 17:15 | ||
17:32
librasteve_ joined
|
|||
Geth | rakudo/release-2025.06.1: 609b90335d | (Justin DeVuyst)++ | 3 files Update changelog + announcement Deliberately not logged: [9847a6ac][51fb0954] |
17:50 | |
lizmat | jdv++ | 18:02 | |
librasteve_ | weekly => rakudoweekly.blog/2025/06/30/2025-06-released/ | 18:11 | |
Geth | nqp/main: 8804e251b5 | (Justin DeVuyst)++ | VERSION [release] Bump VERSION to 2025.06.1 |
18:58 | |
rakudo/release-2025.06.1: f856293c1c | (Justin DeVuyst)++ | tools/templates/NQP_REVISION [release] Bump NQP revision to 2025.06.1 |
|||
rakudo/release-2025.06.1: cf048a948a | (Justin DeVuyst)++ | VERSION [release] Bump VERSION to 2025.06.1 |
|||
rakudo: jdv++ created pull request #5924: Release 2025.06.1 |
19:01 | ||
rakudo/main: 4 commits pushed by (Justin DeVuyst)++ | |||
lizmat | jdv++ again! | 19:04 | |
jdv | patrickb: 2025.06.1 point release done | 19:10 | |
.tell El_Che 2025.06.1 point release done | |||
tellable6 | jdv, I'll pass your message to El_Che | ||
jdv | commit away! | ||
patrickb | Thank you very much! | 19:12 | |
[Coke] | the 0 vs. 01 was a thing I was confused about, don't worry about that. :) | 19:31 | |
[Coke] adds another drink ticket to jdv's tab. | |||
21:40
apogee_ntv joined
21:56
apogee_ntv left,
apogee_ntv joined
21:57
apogee_ntv left
22:09
apogee_ntv joined
|