| [Coke] | Super very low priority, but if commit messages are "RakuAST: Thing" instead of "RakuAST: thing" it saves me a bit of effort. | 00:01 | |
| Geth | rakudo/main: 6 commits pushed by (Nick Logan)++
|
00:12 | |
| rakudo/main: 1386f7bbde | (Nick Logan)++ (committed using GitHub Web editor) | src/Raku/ast/base.rakumod RakuAST: classify the node before offering it to the optimize ladder (#6597) Previously the optimize walk offered every expression to every rewrite and mark in turn. Each helper began by testing the node against the types it accepts, so a typical node paid for thirty method calls that all declined, and the soft scope walk ran for nodes no mark could accept. The optimize stage of the CORE.c build spent 9.9 seconds this ... (8 more lines) |
00:13 | ||
| rakudo: ugexe++ created pull request #6598: Initialize %*ENV, $*CWD and $*SCHEDULER lazily |
00:14 | ||
|
00:50
kurahaupo left,
kurahaupo joined
|
|||
| Geth | rakudo: ugexe++ created pull request #6599: RakuAST: Fix -n/-p loop wrapping issues |
03:57 | |
|
08:05
kurahaupo left
09:49
finanalyst joined
09:57
kurahaupo joined
10:04
kurahaupo left,
kurahaupo joined
11:27
kurahaupo1 joined
11:31
kurahaupo left
13:11
finanalyst left
|
|||
| ab5tract | Would it be relatively straightforward to add a ‘ra:’ chat prefix to the camelia code which would set RAKUDO_RAKUAST=1 for the invocation? | 14:47 | |
| Or is it a little too late to bother with such a convenience? | 14:48 | ||
| ugexe: personally I like the look of that $/ patch. I feel like 6.e is sort of a Raku 2.0, so it would make sense to me personally to take care of this now rather than later in 6.f | 15:08 | ||
| Geth | rakudo: ab5tract++ created pull request #6601: RakuAST: don't lower in a frame whose signature needs the binder |
16:15 | |
| nqp: ugexe++ created pull request #862: Serialize precomputed grammar NFAs in their VM form |
17:46 | ||
| rakudo: ugexe++ created pull request #6602: RakuAST: drop a statement's match tree once its node exists |
17:47 | ||
| rakudo: ugexe++ created pull request #6603: RakuAST: keep a slurpy for body off the statement form |
17:48 | ||
|
18:38
timo left
18:39
timo joined
19:02
kurahaupo1 left
21:06
kurahaupo joined
21:08
[Coke] left
21:33
[Coke] joined
|
|||
| Geth | rakudo/main: 6 commits pushed by (Nick Logan)++
|
21:51 | |
| rakudo/main: 5b01efdd15 | (Nick Logan)++ (committed using GitHub Web editor) | 3 files RakuAST: drop a statement's match tree once its node exists (#6602) Previously every statement's captures stayed reachable for the whole compilation. The statement list only ever reads a statement's node, yet each finished statement kept its capture list and hash, and through them every cursor, capture hash and array beneath it. A closure created while checking the compilation unit pins the frames of the parse, so the tree ... (8 more lines) |
|||
| rakudo/main: c7281e4480 | (Nick Logan)++ (committed using GitHub Web editor) | 2 files RakuAST: keep a slurpy for body off the statement form (#6603) Previously a sunk for loop whose body took a slurpy or capture parameter failed to compile with "This type cannot unbox to a native integer". The count of such a signature is a boxed Inf, and the statement form check compared it to 1 with an NQP ==, which unboxes through .Int. This checks that the count is an Int before comparing, so such a body goes through map like any other body without a fixed count. |
21:52 | ||
|
22:01
finanalyst joined
22:05
finanalyst left
|
|||