|
00:04
sjn left
01:35
[Coke] left
01:37
[Coke] joined
05:43
hurufu joined
09:48
hurufu left
11:01
sjn joined
|
|||
| timo | random observation: env RAKUDO_RAKUAST=1 rakudo -e 'sub test(Num() $x where 1..*) { }; say &test.signature.params[0].constraint_list[0].raku;' → -> ;; Mu $_? = OUTER::<$_> { #`(Block|955247808) ... } | 13:04 | |
| without RAKUDO_RAKUAST: -> ;; Mu $_? = OUTER::<$_> { #`(Block|449520608) ... } | |||
| wrong paste | |||
| -> ;; $_ { #`(Block|1041456880) ... } | |||
| the "use outer $_ as default" in rakuast is probably unnecessary, or perhaps it's even wrong, though I can't think of a way to not pass an argument to the constraint from our own code | 13:05 | ||
| Geth | rakudo/lizmat-9: e2913d03cb | (Elizabeth Mattijsen)++ | src/main.nqp Make RAKUDO_RAKUAST=0 *not* activate the RakUAST logic Spotted by timo++ |
14:02 | |
| rakudo: lizmat++ created pull request #6214: Make RAKUDO_RAKUAST=0 *not* activate the RakUAST logic |
|||
| timo | put you a comment | 14:07 | |
| I was thinking without your change anything is valid in the env var and only env var missing or empty string in it turns rakuast off; with the change I expect RAKUDO_RAKUAST=hello will cause rakudo to crash in versions that have the check and not crash in earlier versions | 14:14 | ||
| actually, it'd be easier if we just string-compare against "" or "0" maybe? | 14:20 | ||
| instead of attempting to intify | |||
| actually looks like i was wrong and +"abcd" just gives 0 so that's actually fine | 14:25 | ||
|
14:26
librasteve_ joined
|
|||
| [Coke] | m: say +"abcd" | 17:09 | |
| camelia | Cannot convert string to number: base-10 number must begin with valid digits or '.' in '<HERE>abcd' (indicated by <HERE>) in block <unit> at <tmp> line 1 |
||
| timo | nqp: say(+"äbcd") | 17:22 | |
| camelia | We trust you have received the usual lecture from the local System Administrator. It usually boils down to these three things: #1) Respect the privacy of others. #2) Think before you type. #3) With great power comes great responsibility… |
||
| timo | uhhh i guess not | ||
| librasteve_ | rakudoweekly.blog/2026/06/01/2026-...ast-fruit/ | 17:25 | |
| lizmat | librasteve_++ | 18:01 | |
| Geth | rakudo/main: 8ec11ecc2f | (Elizabeth Mattijsen)++ | src/core.c/core_epilogue.rakumod Make sure .assuming passes on slurpy hashes In reference to github.com/rakudo/rakudo/issues/6212 Whenever there was a slurpy hash, it would not pass on the slurpy hash argument. That was wrong, it should, and it should actually slip the hash in. ... (5 more lines) |
18:49 | |
| rakudo/we_know_dyncall_plus_clang_is_broken_so_make_ci_green: 6a2787ef19 | (Timo Paulssen)++ | t/04-nativecall/02-simple-args.t TODO both problematic nativecall tests for moar+clang+dyncall the code clang generates assumes the upper bytes of an unsigned char or unsigned short are zeroed out, but dyncall only has a function for passing signed char and short, where negative numbers end up sign-extended, messing clang's generated assembly for x == 0xFE up, which tries to comare the whole 32bit register against 0xFE. |
19:34 | ||
| rakudo: timo++ created pull request #6215: TODO both problematic nativecall tests for moar+clang+dyncall |
|||
| timo | this one's for you [Coke] | ||
| [Coke] | timo: do we want skip or todo? | 19:35 | |
| timo | dunno, i stole the skip from the other test which i assume had the exact same root cause but no "only macos" condition for the skip | 19:36 | |
| lizmat | and now for something completely different: | 19:38 | |
| m: sub a(Int:D $b) { }; dd :type(.type), :modifier(.modifier) given &a.signature.params.head | |||
| camelia | :modifier(":D") :type(Int) |
19:39 | |
| lizmat | so it looks that Parameters don't actually store Int:D as the type, but as Int and some flags | ||
| this breaks this case of .assuming: | 19:41 | ||
| m: sub a($a, Int:D $b) { }; dd &a.assuming(42).signature | |||
| camelia | :(Int $b) | ||
| lizmat | that should have been :(Int:D $b) | 19:42 | |
| m: sub a($a, Int:D $b) { }; dd &a.assuming(42)(Int) | |||
| camelia | Parameter '$b' of routine 'a' must be an object instance of type 'Int', not a type object of type 'Int'. Did you forget a '.new'? in sub a at <tmp> line 1 in block <unit> at <tmp> line 1 |
||
| lizmat | it will still break because of the inner signature, but still | 19:43 | |
| [Coke] | timo++ | 19:44 | |
| timo | well, at least CI will be green now | 20:12 | |
| [Coke] | That will be very helpful. | 20:18 | |
| and now we have in the test a note about what we don't expect to work. | 20:19 | ||
| timo++ | |||
| Geth | ¦ raku.org: librasteve self-assigned Dead "Install" link on Tools page github.com/Raku/raku.org/issues/321 | 21:07 | |
| rakudo: ugexe++ created pull request #6216: RakuAST: allow anonymous declarations across all package kinds |
21:09 | ||
| setup-raku/dependabot/npm_and_yarn/vitest-4.1.0: cb90dc9e2d | dependabot[bot]++ (committed using GitHub Web editor) | 2 files Bump vitest from 4.0.18 to 4.1.0 Bumps [vitest](github.com/vitest-dev/vitest/tree/...es/vitest) from 4.0.18 to 4.1.0. - [Release notes](github.com/vitest-dev/vitest/releases) - [Changelog](github.com/vitest-dev/vitest/blob/...leases.md) - [Commits](github.com/vitest-dev/vitest/commi...es/vitest) ... (8 more lines) |
23:51 | ||
| setup-raku: dependabot[bot]++ created pull request #53: Bump vitest from 4.0.18 to 4.1.0 |
|||