Write organization and language server TODOs

Author: thaliaarchi

omniwsa-ls/src/main.rs

@@ -26,6 +26,11 @@ use serde_json::{from_value as from_json, to_value as to_json};
 // TODO:
 // - Implement text document API, instead of reading from disk.
 // - Record spans in tokens.
+// - About lazy analyses and ties between syntax and semantics:
+//   https://rust-analyzer.github.io/blog/2023/12/26/the-heart-of-a-language-server.html
+//   - To infer imports, it could eagerly parse every Whitespace file. For
+//     inactive files, only labels would need to be retained, instead of the
+//     full CST.
 
 fn main() {
     if let Err(err) = do_main() {

src/dialects/palaiologos/parse.rs

@@ -27,6 +27,7 @@ use crate::{
 //   - Allow using invalid mnemonics as label defs and refs.
 //   - Allow adjacent instructions without slashes. At every misplaced mnemonic,
 //     unless the previous was `rep` or it's used as a label, start a new token.
+//   - Recover `break:` as `label break`, instead of `jmp reak:`.
 
 /// A parser for the Palaiologos Whitespace assembly dialect.
 #[derive(Clone, Debug)]

src/tokens/comment.rs

@@ -6,6 +6,10 @@ use enumset::{EnumSet, EnumSetType};
 
 use crate::syntax::{HasError, Pretty};
 
+// TODO:
+// - Block comments should be parsed into a list or even hierarchy of tokens.
+//   - Line comments within block comments could be highlighted differently.
+
 /// Line comment token (e.g., `#` or `//`).
 #[derive(Clone, DebugCustom, PartialEq, Eq)]
 pub struct LineCommentToken<'s> {

src/tokens/integer/parse2.rs

@@ -6,6 +6,7 @@ use rug::Integer;
 use crate::{lex::Scanner, tokens::integer::BaseStyle};
 
 // TODO:
+// - Parse parens as `GroupToken`.
 // - Implement remainder of parsing, particularly digits.
 // - Handle digit separator position errors.
 // - Handle `0` octal prefix and `0b`/`0B` prefix as digits for `h`/`H` suffix.

src/tokens/string/tokens.rs

@@ -9,10 +9,13 @@ use enumset::{EnumSet, EnumSetType};
 use crate::syntax::{HasError, Pretty};
 
 // TODO:
+// - String tokens could be like `format_args!` and have a list of literal
+//   chunks and escapes, where each escape would have source and parsed
+//   interpretations and its own errors.
 // - Create StringSyntax to describe escapes.
 // - How to represent char literals with buggy delimiters, like those allowed
 //   with littleBugHunter's `'..` pattern? Maybe QuoteStyle::Custom with open
-//   and close.
+//   and close. Or just mark it as unterminated and error.
 // - Improve Debug for CharData::Byte.
 
 /// A string literal token.

src/tokens/token.rs

@@ -22,9 +22,18 @@ use crate::{
 };
 
 // TODO:
-// - Whitelips, Lime, and Respace macro definitions.
-// - Respace `@define`.
 // - Store spans in tokens.
+//   - `Spanned<T>` or `Loc<T>`.
+//   - Use length-based spans, instead of offsets, like Rowan. Since the root
+//     node has the length of the full file, it needs to be `u64`, not shorter.
+// - Organization:
+//   - Merge `Token` into CST.
+//   - Remove Token suffix from type names. What about `StringToken`?
+//   - Possibly reference enum variants as `cst::Variant` like `ty::FnDef(_, _)`
+//     for `rustc_middle::ty::TyKind::FnDef(_, _)`.
+// - Features:
+//   - Whitelips, Lime, and Respace macro definitions.
+//   - Respace `@define`.
 // - Store byte string uniformly (named Text?), instead of a mix of &[u8] and
 //   Cow.
 //   - Create utilities for slicing and manipulating easier than Cow.