I'm thinking that `disjoint` is already a negation: (dis- (not) + joint (united)). When composing with `not`, the user gets a double negation `not (disjoint x y)`. There is a then a small mental effort required to go from "not disjoint" to "overlapping". If we are going to have only one of the two properties, I would rather have the positive one (`overlaps`) as primitive. Then `disjoint` would be written "not (overlaps x y)", which reads quite easily. (Or even "not (x `overlaps` y)"). — Víctor Den 2017-12-19 kl. 16:01, skrev Joachim Breitner:
Hi,
Am Dienstag, den 19.12.2017, 13:44 +0100 schrieb Andreas Abel:
+1 for "disjoint".
+1
I think "overlaps" falls below the Fairbairn threshold.
✓
I always wondered why there is a "notMember" function in the Set interface, saving us 3 key presses.
Probably because of use like this:
filter (`notMember` seen) todo -- pretty
vs.
filter (not . (`member` seen)) todo -- too many parenthesis.
Of course
filter (\x -> not (x `member` seen)) todo -- is also ok
And I will refrain from pointing out that with the idea of no-white- space-means-higher-precedence[1] would allow
filter (not . `member`seen) todo -- too many parenthesis.
[1] https://www.joachim-breitner.de/blog/730-Less_parentheses
One thing to consider: Data.Set should then also be equipped with a function "disjoint", to keep interfaces in sync.
✓
Cheers, Joachim
_______________________________________________ Libraries mailing list Libraries@haskell.org http://mail.haskell.org/cgi-bin/mailman/listinfo/libraries