| Geth | rakudo: ugexe++ created pull request #6529: RakuAST: unify an imported package with its same-named GLOBAL entry |
01:14 | |
| [Coke] | up to 84 fails with ⏳ 1684 out of 2499 modules processed (67.38%) | 02:13 | |
| ... shit. I just fatfingered the tmux connection and killed the blin run | 02:14 | ||
| restarting, will leave it running overnight. | 02:16 | ||
|
05:01
vrurg_ joined
05:04
vrurg left
06:55
finanalyst left
|
|||
| Geth | roast: ec9c85ef0f | (Dan Kogai)++ (committed using GitHub Web editor) | 2 files Add tests for the %a/%A sprintf directives Implemented for the 6.e Formatter by rakudo/rakudo#6525 (with the default sprintf path covered by Raku/nqp#859). Follows the layout of the other per-directive sprintf test files: flag permutations combined with size/precision specifications, tested for both %a and %A. ... (8 more lines) |
07:58 | |
| linkable6 | RAKUDO#6525 [closed]: github.com/rakudo/rakudo/pull/6525 Add %a/%A (C99 hex float) directives to the 6.e Formatter | ||
| linkable6 | NQP#859 [closed]: github.com/Raku/nqp/pull/859 Add %a/%A (C99 hex float) directives to HLL sprintf | 07:59 | |
| Geth | roast: 6da0d49fc3 | (Elizabeth Mattijsen)++ | S32-str/sprintf-a.t Whitespace adjustment for readability Apparently Claude doesn't grok that yet :-) |
08:06 | |
| rakudo/main: 43358324ef | (Elizabeth Mattijsen)++ | src/core.e/Formatter.rakumod Make %a/A format a few percent faster By calling nqp::base_I directly, as the radix is known at compile time, so we can bypass the base sanity test in Int.base. |
08:31 | ||
| ab5tract | ugexe: ah that makes a lot of sense. In that case I guess I’m on the “treat 0 as int in dispatch” | 10:33 | |
| *on the side of | 11:28 | ||
| [Coke] | ah- ugexe, i didn't fatfinger it, I was just changing tabs when it crashed, probably due to the OOM killer (which I wasn't working around anymore since i had so much swap) | 13:58 | |
| so guessing there's some pathological memory issue somewhere. 103 failures so far. | |||
| releasable6: next | 13:59 | ||
| releasable6 | [Coke], Next release in ≈16 days and ≈5 hours. There are no known blockers. 40 out of 69 commits logged | ||
| [Coke], Details: gist.github.com/a9db02ba64e6676a3d...944cad6203 | |||
| ugexe | is there a way to tell what it was processing when it happened? | 14:13 | |
| like once its finished can we compare some list of distributions that failed combined with the list of distributions that passed to find that some distributions aren't listed at all | |||
| looks like recent openssl changes broke things on rakuast | 15:11 | ||
| github.com/raku-community-modules/...7ba707dc0a switched every digest from buf8.allocate to blob8.allocate which breaks PKCS5 and Auth::SCRAM | 15:13 | ||
| ah its broken on legacy frontend as well | 15:18 | ||
| an idea for a tool or add-on for a tool would be for emitting a warning or something when trying to release something that changes the return type of an existing function without bumping the major version | 15:23 | ||
| lizmat | ugexe: will fix OpenSSL | 16:41 | |
| hmmm... how come this wasn't spotted before: that means the problem in OpenSSL for these modules existed for more than a year ? | 16:48 | ||
| ugexe | probably why the modules i mentioned are marked AlwaysFail on blin | 16:52 | |
| lizmat | github.com/raku-community-modules/...ba97d085ce # on its way to the ecosystem now | 17:08 | |
|
17:16
Sp1983 joined
17:20
Sp1983 left
|
|||
| [Coke] | ugexe: (what it was processing) - when I'm not using the system manager, it just kills the entire tmux tab. I have to add that back in, and then maybe I can guess based on the last X items on the logged output | 17:20 | |
| Yah, we've only been looking at blin for things that are NEW fails. | 17:21 | ||
| if it fails old and new, it's not mentioned. (and it's possible the module itself could be responsible for something that went from "works on both" checking in release A, to "fails on both" on releae A+1 | |||
| For things marked alwaysfail - My goal is to curate a list of things that we are ok to have as AlwaysFail (basically, unmaintained items); and then we can keep a sharper eye on what's left. | 17:22 | ||
| lizmat | well, I hope we're going to have 2 less alwaysfails now :-) | 17:23 | |
| Geth | Blin/coke/cache-listings: 86d6dcc055 | (Will Coleda)++ | bin/blin.p6 WIP |
17:46 | |
| Blin/coke/cleanup: 63e7fc0efa | (Will Coleda)++ | bin/blin.p6 fix var name |
|||
| [Coke] | Could geth announce branch removals? | 19:07 | |
| (and should it if it can?) | |||
| lizmat | it will be | 19:15 | |
| Geth | Blin/coke/cleanup: 17 commits pushed by (Will Coleda)++ review: github.com/Raku/Blin/compare/63e7f...b8d9e0f740 |
19:18 | |
| ugexe | github.com/ugexe/Inline-Perl5/comm...7b9e41d391 | 19:58 | |
| lizmat | ugexe: thanks | 20:09 | |
| ugexe | maybe that implies there should be an $*R.add_phaser though | ||
| lizmat | $resolver.find-attach-target('block').add-phaser("QUIT", self); ? | 20:10 | |
| feels like that should be easy to make then ? | 20:11 | ||
| ugexe | github.com/ugexe/Inline-Perl5/blob...5-inline.t is the test that doesn't pass yet | ||
| lizmat | from ast/statementprefixes.rakumod 909 | ||
| afk& | |||
|
21:36
coleman left
21:40
coleman joined
|
|||
| Geth | rakudo/main: 5 commits pushed by (Nick Logan)++
|
21:50 | |
| ugexe | like ~20 of the blin failures were a regression from the last blin run about a redeclared symbol. that merge should fix all of those | 21:51 | |
| Geth | rakudo/main: 4b4af55da3 | (Nick Logan)++ (committed using GitHub Web editor) | 7 files RakuAST: never fold a constant-string indirect lookup (#6528) Previously an indirect lookup written with a constant string, like ::("EXPORT") or ::("Foo"), resolved during compilation. Consumers of compile-time values then folded the parse-time symbol, and a .&::("foo") call compiled a direct call to its parse-time resolution. The runtime symbolic lookup consults the symbols a require merged into the caller's ... (38 more lines) |
21:52 | |
| [Coke] | github.com/rakudo/rakudo/compare/4...68024ad53d from the Geth line above is broken. "There isn’t anything to compare. | 22:40 | |