Geth rakudo: ugexe++ created pull request #6203:
RakuAST: smartmatch against a Match no longer returns a magic value
02:13
releasable6 Next release in ≈1 day and ≈15 hours. There are no known blockers. Please log your changes in the ChangeLog: github.com/rakudo/rakudo/wiki/ChangeLog-Draft 03:00
05:07 hurufu joined 06:11 hurufu left 06:12 hurufu joined 11:31 donaldh left, patrickb left, vrurg_ left 11:35 donaldh joined, patrickb joined, vrurg_ joined 11:51 hurufu left 12:12 hurufu joined 12:37 hurufu left 12:43 hurufu joined
[Coke] So, I happen to have a day off from work today. I don't suppose anyone would be mad if the release happened a day early? In any case, I'll get it started probably around noon here. 13:16
lizmat [Coke]++ 13:24
and +1
I just realized that stringification of FatRats has been limited to 6 digits since like forever 13:25
and I guess I also want to change that to 16 digits... but that would require spectest changes... which I'm more comfortable with now, then 7 years ago when I first noticed that oddity 13:26
also, I'm thinking of making stringification of (Fat)Rats to be special cases of the sprintf %g format, rather than the other way aroudn 13:27
disbot4 <librasteve> not sure what you get from the additional 10 digits (16-6) ... from a stringification, I guess you want (i) a reasonably readable format, (ii) an idea of the value, different strings are always different values and (iv) an idea of the type 14:48
<librasteve> the score for 16 vs 6 is (i) worse (16 bits is "too much info"), (ii) worse (ditto), (iii) same (can't ever guarantee for (Fat)Rats) and (iv) same 14:50
<librasteve> to me it boils down to do you want you FatRat strings to look different to Rat strings - personally I am fine with them looking the same and calling dd or ^name to work out what I got 14:51
<librasteve> anyway use FatRatStr and get the whole hog!
15:47 thundergnat joined
thundergnat FWIW, I would be in favor of increasing the default minimum precision of FatRats. When I submitted the pull request overhauling and speeding up FatRat stringification 7(?) years ago, this comment was added. github.com/rakudo/rakudo/blob/main...kumod#L130  lizmat ended up doing further optimizing before it was merged, 15:51
but the comment is still there.
When it didn't happen, I wrote raku.land/zef:thundergnat/Rat::Precise to make up for it. (Which is more configurable anyway.) 15:52
Geth rakudo/release-2026.05: 4770cbe182 | (Will Coleda)++ | 3 files
Update changelog + announcement

Deliberately not logged:
  [4acea051][eb798378][f6d89e38][a1f6cb4f][4f641591][0523e6b2][155528fe]
  [ed5ddf8c][300ee1d9][f2273bf1]
thundergnat My point being: thumbs up from me. (Not a dev, but a frequent user of Raku.) 15:53
15:53 thundergnat left
[Coke] MoarVM release in progress 16:05
timo wheee 16:06
16:48 vrurg joined 16:51 vrurg_ left 18:29 hurufu left
[Coke] sorry, left for lunch half way through. :) 18:44
MoarVM 2026.05 done 19:19
please fix the .perl call in JSON::Fast. It's also showing up in the release process. 19:38
patrickb I might manage to do the binaries tonight. But I still can't do Mac on arm. My Mac stadium machine has turned inaccessible for some reason... 19:56
timo: Did you also have access to that machine? Can you try if you can still login? 19:57
[Coke] So many errata 6,c failures. 20:24
... and the 6.d failures from last month still open. 20:25
github.com/Raku/roast/issues/886 20:26
github.com/Raku/roast/issues/888
trying 6.d now... 20:27
Do we have any CI for roast? 20:29
I wouldn't expect it on every release, but would be nice to run all language versions once a week to avoid surprises. 20:30
sorry, I wouldn't expect it on every *commit*
github.com/Raku/roast/issues/889 20:40
failures in set/bag/mix/currying
looks like SBM might just be "copy changes from master" 20:41
curry tests haven't been touched on master since 2025, so that might be an actual dev issue. 20:42
going to pause the 2026.05 release, ask if any devs can double check those 3 tickets.
good news - we're a day early, and even if this needs a dev fix, we're still fine for tomorrow. 20:43
ugexe We do run CI roast tests for every commit, with and without rakuast. I guess make spectest isnt running the errata then 21:08
patrickb I regained access. 2026.04 Mac on arm binary is uploaded. \o/ 21:12
ugexe Just fyi I won’t be at a computer much this weekend (including today). It looks like the currying tests that are failing are using something called Test::Assuming 21:18
lizmat [Coke] I'm trying to repro github.com/Raku/roast/issues/886 but it all passes for me? 21:21
what am I doing wrong?
[Coke] lizmat: that was from last month - if it's not in 888, we can close it 21:31
sorry for the noise 21:32
lizmat the tests in t/spec/S06-currying/named.t are wrong
[Coke] 888 is the fresh 6.d run
lizmat m: sub a(:$a, :$b) { }; my &b = &a.assuming(:a<foo>); dd &a.signature, &b.signature
evalable6 :(:$a, :$b)
:(:$a = "foo", :$b)
lizmat the tests claim that add an assuming with a named arg, doesn't change the signature.
but that is wrong, so what shall I do with the tests ? 21:33
because they enshrined incorrect behaviour
[Coke] my 2¢ is to fix the tests with a commit message that indicates that these tests were incorrect and we're fixing it now. 21:34
but let's get at least one other person to agree to that.
Do you think there will be a rakudo change to get past this, or all test updates? 21:35
Just wondering if I have to redo any of release so far or just rerun the tests
21:36 hurufu joined
Geth roast/6.d-errata: a429ce1c18 | (Elizabeth Mattijsen)++ | 2 files
Relax immutability tests

After the Great QuantHash Refactor the error type changed when trying to mutate a Bag/Mix. Check for generic execution error rather than for the specific type.
21:37
22:09 hurufu left
Geth roast/6.d-errata: 675fd00ab4 | (Elizabeth Mattijsen)++ | 3 files
Skip all bad currying/priming tests

The old .assuming implementation had many issues, and many of them were enshrined in spectests :-(
22:33
roast/6.d-errata: 092fe351f5 | (Elizabeth Mattijsen)++ | S32-basics/xxKEY.t
Another set of immutability tests relaxations
22:37
lizmat [Coke]: something weird going on with t/spec/S03-operators/set_difference.t that I'm not seeing atm 22:46
and it's late here... will look tomorrow
I think I've fixed all of the other failing tests 22:49
releasable6 Next release in ≈19 hours. GitHub is down. There are no known blockers. Please log your changes in the ChangeLog: github.com/rakudo/rakudo/wiki/ChangeLog-Draft 23:00
[Coke] Whatever changes on 6.d may also need to go on 6.c 23:09
how long has it been since we've had any flapping spectests? 23:13
I am tempted to remove any flap protection from the release.
Geth ¦ rakudo: coke self-assigned Remove any flapper "protection" from the release process github.com/rakudo/rakudo/issues/6205 23:17
[Coke] releasable6: next 23:19
releasable6 [Coke], Next release in ≈19 hours. There are no known blockers. 139 out of 139 commits logged