Implement private modules #7840
Conversation
Akirathan
force-pushed
the
wip/akirathan/7304-Implement-private-modules
branch
from
3b2e5e5 to
08513fe
Compare
lib/rust/parser/src/syntax/tree.rs
Outdated
| @@ -951,6 +955,7 @@ pub fn apply_unary_operator<'s>(opr: token::Operator<'s>, rhs: Option<Tree<'s>>) | |||
| pub fn to_ast(token: Token) -> Tree { | |||
| match token.variant { | |||
| token::Variant::Ident(ident) => token.with_variant(ident).into(), | |||
| token::Variant::Private(private) => Tree::private(token.with_variant(private)), | |||
There was a problem hiding this comment.
This will never match because the Private token type is only generated by the macro, which immediately places it in a AST node, so an Item::Token(Token::Private) is never encountered.
I don't see a need for a token type for private; if it were a reserved word we could easily identify at lex time and always treat specially then we should give it a token type, but we are taking a contextual keywords approach (because it is a goal that identifiers with special meanings in syntactic constructs can be used as normal identifiers when used outside those constructs).
| println!("private_keyword[begin] : Segments={:?}", segments); | ||
| let segment = segments.pop().0; | ||
| let keyword = into_private(segment.header); | ||
| let ret_tree = syntax::Tree::private(keyword); |
There was a problem hiding this comment.
The body can be handled the same way as in capture_expressions, and placed in an Option<Tree> field of the private node. (I think an optional body is preferable to separate Tree types for the standalone-private and private-with-expression because we have to assume the user could be halfway through typing something, so a Private without a body may well be private-with-expression whose expression hasn't been written yet).
|
|
||
| #[test] | ||
| #[ignore] | ||
| fn private_is_first_statement() { |
There was a problem hiding this comment.
This will require changes to the macro resolver--we could add a field to Context::Statement, like so:
enum Context { Expression, Statement { first: bool } }
The resolver would then keep track of when the first "contentful' line has been encountered in order to construct the first field. Then we could add a third MacroMap that is only run in Statement context with first=true. Note that there's a rewrite of the macro resolver currently in CI: #7711.
| #[rustfmt::skip] | ||
| let lines = vec
Co-authored-by: Jaroslav Tulach <[email protected]>
….com:enso-org/enso into wip/akirathan/7304-Implement-private-modules
Akirathan
added
CI: Clean build required
Akirathan
removed
the
CI: Keep up to date
| if (subModName.getParent().isDefined()) { | ||
| return parentModName.item().equals( | ||
| subModName.getParent().get().item() | ||
| ); | ||
| } else { | ||
| return false; | ||
| } |
| } else { | ||
| throw new IllegalStateException("unknown exp: " + exp); | ||
| } | ||
| return null; |
There was a problem hiding this comment.
is this correct? Shouldn't this be return false;?
Closes #7304
Pull Request Description
Adds the ability to declare a module as private. Modifies the parser to add the
privatekeyword as a reserved keyword. All the checks for private modules are implemented as an independent Compiler pass. No checks are done at runtime.Important Notes
private- a reserved keyword.privatekeyword as the first statement are declared as private (Project private)Checklist
Please ensure that the following checklist has been satisfied before submitting the PR:
Scala,
Java,
and
Rust
style guides. In case you are using a language not listed above, follow the Rust style guide.
./run ide build.