Follow-up after #6805.
`TraverseCtx::generate_binding` take an `Atom` instead of a `CompactStr`. If it's an existing var name (which it must be, otherwise we'd be using `TraverseCtx::generate_uid`), an `Atom` will already exist in the arena for this symbol name, so we'd be better off reusing it, rather than writing the same `Atom` into the arena again.
In particular:
* the long comment was scrolling off side of screen on
[npm.com](https://www.npmjs.com/package/oxc-parser).
* the example is ESM, so can simplify it by using top level await.
These two APIs are used to create a symbol with the provided symbol name, The difference from `generate_uid` is that we don't need to create a unique name for the symbol.
If you have a better method name for this, Feel free to edit this PR directly
Style nit. Dereference `&ScopeId` to `ScopeId` as early as possible. `&ScopeId` is 8 bytes, whereas `ScopeId` is 4 bytes.
In simple cases like this, compiler will optimize it anyway, but still I think it's a better pattern to dererence early. In more complicated cases, it will be better for performance, and in my opinion, it makes things clearer if vars called `scope_id` are always a `ScopeId`, not sometimes a `&ScopeId`.
Store `BoundIdentifier` in `registrations`. This allows removing the call to `ctx.ast.atom()` which re-allocates a new `Atom` into the arena - unnecessary as it's already been allocated when the UID was created.
It's unfortunate that `Semantic` stores symbol names as `CompactStr`s instead of `Atom`s, otherwise we could get the name as an `Atom` without re-allocating it. Hopefully we can make that change in future, and then reduce `BoundIdentifier` to just contain the `SymbolId`. But for now, we are kind of stuck with it.
Style nit. Dereference `&SymbolId` to `SymbolId` as early as possible. `&SymbolId` is 8 bytes, whereas `SymbolId` is 4 bytes.
In simple cases like these, compiler will optimize it anyway, but still I think it's a better pattern to dererence early. In more complicated cases, it will be better for performance, and in my opinion, it makes things clearer if vars called `symbol_id` are always a `SymbolId`, and not sometimes a `&SymbolId`.
Style nit. Dereference `&ScopeId` to `ScopeId` (and other IDs) as early as possible. `&ScopeId` is 8 bytes, whereas `ScopeId` is 4 bytes.
In simple cases like these, compiler will optimize it anyway, but still I think it's a better pattern to dererence early. In more complicated cases, it will be better for performance, and in my opinion, it makes things clearer if vars called `scope_id` are always a `ScopeId`, not sometimes a `&ScopeId`.
First of a series of PRs removing `new` and `new_with_scope_id` etc methods from AST types. Following #6760, all AST node creation can now go via the AST builder.
This lays groundwork for Node IDs (#5689), as we'll need `NodeId`s to be generated in `AstBuilder`, so that all nodes receive an ID.
Continue work on #4742.
Only `oxlint --print-config all` is supported. It's useful to migrate from command-line interface to config file.
The `--print-config PATH` looks not really useful for us now, I will add it after config file overrides supported.
