Labeled block not allowed as lhs of assignment at statement level #4

Hejsil commented 2019年03月18日 15:54:00 +01:00 (Migrated from github.com)

Explained here

Actually, this broke with the grammar changes in #1685. Basically, we have this rule at stmt level:

Statement
 <- ...
 / LabeledStatement
 ...
 / AssignExpr SEMICOLON
LabeledStatement <- BlockLabel? (Block / LoopStatement)

AssignExpr expands into all other expressions. LabeledStatement is before AssignExpr so that rule is being parsed first, succeeds and then Statement succeeds. Then the parser tries to end the __case_1 block, which then fails the parsing.

The parser parses the grammar correctly, but the new grammar just does not allow this syntax. I can't come up with a solution on top of my head right now.

Edit: Well, a solution would be to make translate-c conform to this new grammar. Idk if we want to put in the work to make x:{}.* = 1; valid at stmt level.

Explained [here](https://github.com/ziglang/zig/issues/2043#issuecomment-472900431) > Actually, this broke with the grammar changes in #1685. Basically, we have this rule at stmt level: > > ``` > Statement > <- ... > / LabeledStatement > ... > / AssignExpr SEMICOLON > > LabeledStatement <- BlockLabel? (Block / LoopStatement) > ``` > `AssignExpr` expands into all other expressions. `LabeledStatement` is before `AssignExpr` so that rule is being parsed first, succeeds and then `Statement` succeeds. Then the parser tries to end the `__case_1` block, which then fails the parsing. > > The parser parses the grammar correctly, but the new grammar just does not allow this syntax. I can't come up with a solution on top of my head right now. > > Edit: Well, a solution would be to make translate-c conform to this new grammar. Idk if we want to put in the work to make `x:{}.* = 1;` valid at stmt level.