Automated Resyntax fixes #1463
Automated Resyntax fixes #1463
Conversation
![resyntax-ci[bot]](https://avatars.githubusercontent.com/in/999945?s=60&v=4){kind=small}
Specify a line mode of `'any` with `read-line` to avoid differences between Windows and other platforms.
Internal definitions are recommended instead of `let` expressions, to reduce nesting.
This use of `format` does nothing.
Using `cond` instead of `if` here makes `begin` unnecessary
This use of `define-values` is unnecessary.
This `match` expression can be simplified using `match-define`.
Using `when` and `unless` is simpler than a conditional with an always-throwing branch.
`cond` with internal definitions is preferred over `if` with `let`, to reduce nesting
Internal definitions are recommended instead of `let` expressions, to reduce nesting.
This `if`-`else` chain can be converted to a `cond` expression.
The `define` form supports a shorthand for defining functions.
This `if` expression can be refactored to an equivalent expression using `and`.
This list-constructing expression can be simplified
This `let` expression can be pulled up into a `define` expression.
| @@ -915,7 +915,8 @@ | |||
| [(? variance:const?) S] | |||
| [(? variance:co?) S] | |||
| [(? variance:contra?) T] | |||
| [(? variance:inv?) (let ([gS (generalize S)]) (if (subtype gS T) gS S))])) | |||
| [(? variance:inv?) (define gS (generalize S)) | |||
There was a problem hiding this comment.
Is the newline here because of the define or because of the overall line length? @jackfirth
There was a problem hiding this comment.
Ah. This is a bug in Resyntax's efforts to avoid reformatting unchanged code. It should have produced this:
[(? variance:inv?)
(define gs (generalize S))
(if (subtype gS T) gS S)]
| ;; +/- (3 args) | ||
| [((~var e1 (w/obj+type -Int)) (~var e2 (w/obj+type -Int)) (~var e3 (w/obj+type -Int))) | ||
| (define (sign o) | ||
| (if plus? |
There was a problem hiding this comment.
There's more context in sorawee/fmt#75, but the gist is that a lot of folks find one-line if expressions difficult to read even when they're short. The current rule that fmt uses is to only allow one-line ifs if the true and false branches are atoms.
This one does seem a bit unnecessary. I'd like to get more data before suggesting changes to fmt's handling of these, however. Could you keep your eyes out for other cases like this?
There was a problem hiding this comment.
There are a few others in this PR, all of which I dislike. In general I think fmt ended up in a place where it prefers adding a lot more newlines than I like overall (here, in define, in function application) which makes resyntax PRs harder to be happy about.
Resyntax fixed 33 issues in 20 files.
define-lambda-to-definelet-to-defineread-line-anynested-if-to-condalways-throwing-if-to-whenconsing-onto-static-listif-begin-to-condif-else-false-to-andsingle-clause-match-to-match-defineif-let-to-conddefine-values-values-to-definecond-let-to-cond-definedefine-let-to-double-defineformat-identity