|
01:34
kjp left
01:36
kjp joined
02:51
coleman left,
shareable6 left,
zostay left
03:59
coleman joined,
shareable6 joined,
zostay joined
06:47
vrurg_ joined,
vrurg left
06:59
hurufu joined
07:22
vrurg_ left
07:24
vrurg_ joined
|
|||
| ShimmerFairy | I'm curious if there's a description anywhere of the changes that 6.e will bring. In particular, I'm interested in Grammar and what looks to be added error reporting support; it'd be nice if I didn't need an external module for that anymore. | 10:11 | |
| lizmat | I started with an overview., doesn't include 2026.09: github.com/rakudo/rakudo/blob/main...verview.md | 10:16 | |
| ShimmerFairy | Thanks, that'll be handy to keep on hand. With a little bit of playing around, it looks like Grammar's 6.e error reporting isn't quite displaying the right snippet of input text on failure, but I've only just started poking at it. | 10:27 | |
| Geth | rakudo/main: 8cf84d54cd | (Elizabeth Mattijsen)++ | docs/6.e-release-overview.md Add 2026.09 to 6.e release overview Also make the release headers consistent |
10:40 | |
| lizmat | [Coke] ab5tract I didn't think this would need a PR, sorry | ||
| ShimmerFairy | m: use v6.e.PREVIEW; grammar G { token TOP { [ABC\n]+ } }; say G.parse("ABC\nABC\nXYZ\n").exception # I'd expect this to put the eject symbol in front of "XYZ". Is this a bug or working as intended? | 10:41 | |
| camelia | ===SORRY!=== Error while compiling <anon> Confused at <anon>:1 ------> <BOL><HERE>ABC |
||
| lizmat | feels to me that it is just producing the error at pos 0, so probably not all that useful that way ? | 10:45 | |
| ShimmerFairy | Yeah, I'm not sure if there's a more complicated grammar setup that would get it to print a more useful position. I tried with at least TOP calling another rule, and got the same result. | 10:46 | |
| lizmat | m: use v6.e.PREVIEW; grammar G { token TOP { [ABC\n]+ } }; say G.subparse("ABC\nABC\nXYZ") | ||
| camelia | ｢ABC ABC ｣ |
||
| lizmat | feels to me it should at least put the eject symbol at the *end* of that partial match ? | 10:47 | |
| ShimmerFairy | Same here. It feels like either the wrong cursor/match is being sent to the error handling process, or that process isn't using the right position attribute to find where things failed. | 10:49 | |
| (in general, I'd expect this failure mechanism to point to the highwater mark, since under the principle of LTM the part of the grammar that got the farthest is probably the most correct) | 10:51 | ||
| Geth | rakudo/lizmat-129: 2ca5808158 | (Elizabeth Mattijsen)++ | src/core.c/IO/Path.rakumod Further strictness on IO::Path.resolve Since the $!CWD of an IO::Path can really be anything, and an absolute path in a symlink can also be anything really, both will need to be checked (instead of assumed to be ok). So instead of including the $!CWD and an absolute link to the resolved part, add them to the unresolved part. Also add some more internal documentation |
11:46 | |
| rakudo: lizmat++ created pull request #6927: Further strictness on IO::Path.resolve |
|||
| rakudo/lizmat-130: ebbfc15caf | (Elizabeth Mattijsen)++ | src/Raku/Actions.nqp RakuAST: make "use p5isms/c++isms" default in 6.e+ In response to github.com/rakudo/rakudo/issues/6926 So this will allow "say" without any params to work in 6.e, e.g. when being used in a feed operator. I think the chance of someone inadvertently getting caught by these compilation checks, is greater than the confusion a seasoned Perl developer would get by a bare "say" not saying the topic of the calling context. |
12:58 | ||
| rakudo: lizmat++ created pull request #6928: RakuAST: make "use p5isms/c++isms" default in 6.e+ |
13:00 | ||
| rakudo/lizmat-130: ee9f837206 | (Elizabeth Mattijsen)++ | src/Raku/Actions.nqp RakuAST: make "use p5isms/c++isms" default in 6.e+ In response to github.com/rakudo/rakudo/issues/6926 So this will allow "say" without any params to work in 6.e, e.g. when being used in a feed operator. I think the chance of someone inadvertently getting caught by these compilation checks, is greater than the confusion a seasoned Perl developer would get by a bare "say" not saying the topic of the calling context. |
13:01 | ||
| lizmat | sorry for the noise, was initially based on 129 | ||
| Geth | rakudo/lizmat-129: 10e621a5e8 | (Elizabeth Mattijsen)++ | src/core.c/IO/Path.rakumod Make looping symlinks fail Spotted by uniejo++ |
13:26 | |
| ab5tract | I thought all feed sections needed to be a call? | 14:09 | |
| Ah, not the last one | |||