You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
This addresses a regression in commit 8d9cb49.
That commit uses (when (andmap not errs) (expected-but-got ...)) to
suppress some errors when type-checking list elements. Before this,
there would be two errors reported: one for the list element itself
and one for the surrounding list. However, as per #1379 this change
suppressed too much. In certain cases it wouldn't report any errors
for an invalid list.
I couldn't figure out a way to report errors perfectly, so instead,
I decided to revert the relevant change in 8d9cb49. This means list
could report multiple errors again, but it's honestly not that bad.
It shows the innermost location where the problem happened, as well
as the wider context of the list, which can be helpful for locating
and understanding the error. An example using the code from #1379:
sample.rkt:13:41: Type Checker: type mismatch
expected: (Listof Xexpr)
given: (List Positive-Float-No-NaN)
in: (list invalid)
- snip -
sample.rkt:9:2: Type Checker: type mismatch
expected: Xexpr
given: (List 'html (List 'body (List 'p String) (Pairof 'p Nothing)))
in: (quasiquote (html (body (p (unquote (~a invalid))) (p (unquote-splicing ((inst map Xexpr Xexpr) values (list invalid)))))))
- snip -
Type Checker: Summary: 2 errors encountered
location...:
sample.rkt:13:41
sample.rkt:9:2
- snip -
The performance fix from 8d9cb49 still works.
0 commit comments