jackfirth
Remove `apply-append-for-loop-to-for-loop` lint
The apply-append-for-loop-to-for-loop lint seems to, in some cases, result in much uglier code. For example, in Herbie, it rewrites:
,@(apply append
(for/list ([rows (append mainline-infos other-infos)])
(print-rows rows #:name (dict-ref (first rows) 'branch)))
into
,@(for*/list ([rows (append mainline-infos other-infos)]
[v (in-list (print-rows rows #:name (dict-ref (first rows) 'branch)))])
v)
This seems much worse! The actual body of the loop is now hidden inside a loop bound, and there's a dummy variable introduced (v) with no real purpose. It also moves the code way to the right, which makes it uglier too. I suggest removing or scoping down this lint.
I added this one back when I was trying to get a bunch of gnarly list-processing code to fuse into nice loops, so the real value wasn't this specific refactoring - it was combining it with other for-based refactorings. But on its own I agree that it's actually detrimental. I'll remove this one.
Someday I'd like to see the for/append-list macro suggested in racket/racket#2483 implemented. That's nicer than either existing solution, and would let Resyntax rewrite the above code to this:
,@(for/append-list ([rows (append mainline-infos other-infos)])
(print-rows rows #:name (dict-ref (first rows) 'branch)))
Yeah that would be nice. In Herbie we sometimes use a reap macro for this, and in another project I had a for/reap that made stuff like this nice. Whatever, anyway, I filed the issue, removing the lint seems good, or maybe it can be used only if it combines with another lint later or something.