feat(fmt): rewrite formatter using Solar and a structured algorithm #10907
Conversation
| */ | ||
| constructor(string memory name_, string memory symbol_) { | ||
| _name = name_; | ||
| _symbol = symbol_; | ||
| } | ||
|
|
||
| /** | ||
| * @dev See {IERC721-balanceOf}. | ||
| * @dev See {IERC721-balanceOf}. |
There was a problem hiding this comment.
hmm this is kinda weird, i dont think we want to touch indentation inside of comments?
There was a problem hiding this comment.
as you know, for doc comments we must always add the * for new lines.
in the old impl, when adding it, they would add a tab, which made things inconsistent when some lines had * and in other it was added by the formatter.
personally i'd either ensure that there is always a tab (like i did) or completely get rid of tabs in commments
| (10, 20, 30); | ||
| (uint256 listItem004, uint256 listItem005, uint256 listItem006) = ( | ||
| 10, 20, 30 | ||
| ); |
There was a problem hiding this comment.
hmm is this right? shouldn't each the tuple element go on a new line? I don't mind it, just checking
There was a problem hiding this comment.
we support both depending on whether the selected style is Consistent (everything breaks) or Compact (tries to isolate all items on the same line and if they don't fit everything breaks).
I can change it if we want to
|
I've looked at all the tests that are in the diff and LGTM, you can mark the "STYLE" in the description as completed |
| @@ -2,20 +2,20 @@ | |||
| contract A { | |||
| Counter public counter; | |||
| /** | |||
| * TODO: this fuzz use too much time to execute | |||
| * function testGetFuzz(bytes[2][] memory kvs) public { | |||
There was a problem hiding this comment.
@DaniPopes this is an example of a block cmnt where they would add a *\t but which is inconsistent with the user-defined * TODO
that's why i chose to add tabs everywhere
Motivation
Rewrite forge-fmt entirely to use Solar as the Solidity/Yul parser and AST, and using a structured pretty-printing algorithm.
The language-agnostic implementation of the pretty-printer (
src/pp) is based onrustc_ast_prettyandprettyplease, both MIT and Apache licensed. Seeprettyplease's README for algorithm notes.The goal of this PR is to make progress with #9317, and fixing existing issues with the formatter while trying to maintain compatibility as much as possible.
This PR will not be replacing the formatter yet, opting to create a separate crate
crates/fmt-2to reduce unrelated changes and aid with review. A follow-up PR will replacecrates/fmtand deal with fallout in-tree.Issues
Issues that will be fixed with the new implementation. Not closed by this PR, but by the follow-up that will hook the new implementation up; see paragraph above.
line_lengthto be a hard requirement #4450forge fmt): sorting of imports adds newlines in groups #7944forge fmt): ignoresforgefmt: disable-next-linein argument lists #8557transientkeyword by moving to Solar #9088forgefmt: disable-next-line#11249Tests state
Supported Configs:
indent_style = IndentStylemultiline_func_header: MultilineFuncHeaderStyleline_length: usizetab_width: usizebracket_spacing: boolint_types: IntTypesquote_style: QuoteStylenumber_underscore: NumberUnderscorehex_underscore: HexUnderscoresingle_line_statement_blocks: SingleLineBlockStyleoverride_spacing: boolwrap_comments: boolignore: Vec<String>contract_new_lines: boolsort_imports: boolno_space_power: boolfeat(fmt): option to remove spaces between number and exponent #4512