Parse arbitrarily complex box patterns.
#1733
Conversation
|
r? @matklad |
ecstatic-morse
force-pushed
the
nested-box-pattern
branch
2 times, most recently
from
26bba51 to
075348a
Compare
| @@ -416,6 +416,7 @@ pub(crate) fn match_arm_list(p: &mut Parser) { | |||
| // | X => (), | |||
| // box X => (), | |||
| // Some(box X) => (), | |||
| // Some(box Test{field: 0}) => (), | |||
There was a problem hiding this comment.
Hm, actually, I think these last three lines are misplaced. I specifically looked for this in the previous PR and somehow still missed it =/
The purpose of these inline tests is to show what exactly each specific bit of code parses. box patterns are not handled in this function.
I see that you've added tests to fn box_pat, where they belong, so these last three lines should be removed.
| Some(m) | ||
| } | ||
|
|
||
| fn is_literal_pat_start(p: &mut Parser) -> bool { | ||
| fn is_literal_pat_start(p: &Parser) -> bool { |
| // j: Box<Inner<'a>>, | ||
| // } | ||
| // | ||
| // struct Inner<'a>(&'a i32); |
There was a problem hiding this comment.
Let's remove struct Ounter and struct Inner, they are completely irrelvent for testing syntax
There was a problem hiding this comment.
So it seems that this test only passes if there's something named Inner and Outer in scope. This is probably a shortcoming of the parsing code?
There was a problem hiding this comment.
It shouldn't be the case. Judging by the actual .rs file in the inline folder, looks like the test harness is confused over a blank line?
There was a problem hiding this comment.
I get the following error if the names aren't in scope. It's an error in my parsing code, not sure why it went away when I changed the test, maybe the blank line?
thread 'tests::parser_tests' panicked at 'assertion failed: `(left == right)`
left: `[SyntaxError { kind: ParseError(ParseError("expected a name")), location: 52 }, SyntaxError { kind: ParseError(ParseError("expected COMMA")), location: 55 }]`,
right: `[]`: There should be no errors in the file "/home/mackendy/src/rust/rust-analyzer/crates/ra_syntax/test_data/parser/inline/ok/0143_box_pat.rs"', crates/ra_syntax/src/tests.rs:23:9
0143_box_pat.rs:
fn main() {
let box i = ();
let box Outer { box i, j: box Inner(box &x) } = ();
let box ref mut i = ();
}There was a problem hiding this comment.
I needed to extend record_field_pat_list to handle the box i part. This syntax is equivalent to i: box i. I was a little surprised that it is in fact valid.
ecstatic-morse
force-pushed
the
nested-box-pattern
branch
2 times, most recently
from
19d5c9e to
75192f7
Compare
|
@matklad. Okay, this should be good to go now. Do you want me to rebase to clean up the latest changes or does bors squash automatically? |
|
bors doesn't squash, so a rebase wouldn't harm (though I personally don't value clean git history that much for small PR). |
Named structs can have `box` patterns that will bind to their fields. This is similar to the behavior of the `ref` and `mut` fields, but is at least a little bit surprising.
ecstatic-morse
force-pushed
the
nested-box-pattern
branch
from
da0b860 to
c93903e
Compare
|
@matklad I cleaned it up. It's ready to merge if you want to give it one last look. |
|
bors r+ Thanks! |
1733: Parse arbitrarily complex `box` patterns. r=matklad a=ecstatic-morse This fully resolves the pattern part of #1412 by enabling the parsing of complex `box` patterns such as: ```rust let box Struct { box i, j: box Inner(box &x) } = todo!(); ``` This introduces a new `ast::BoxPat` (in the mold of `ast::RefPat`) that gets translated to `hir::Pat::Missing`. Co-authored-by: Dylan MacKenzie <ecstaticmorse@gmail.com>
Build succeeded |
This fully resolves the pattern part of #1412 by enabling the parsing of complex
boxpatterns such as:This introduces a new
ast::BoxPat(in the mold ofast::RefPat) that gets translated tohir::Pat::Missing.