The library coding standards[1] say: * Use '?' suffix for predicates. - N.B. - predicates return booleans
and the community Clojure style guide[2] says:
* The names of predicate methods (methods that return a boolean value) should
end in a question mark. (i.e.even?).
Both of these imply that if you have a function that returns a boolean (and
that is intended for use as a predicate), it should be named to end in '?'.
Fair enough.
My question is about the reverse implication:
* Should a function whose name ends in '?' return a (strict) boolean value?
Looking at the docstrings of a random selection of functions found by (apropos
"?"), they all seem to return specifically true or false. I did not do an
exhaustive check.
Is the intent that foo? implies a result of true or false - or could foo?
return any truthy / falsey value (and therefore any Clojure value).
Concrete example that spurred this discussion from some code at work:
(defn is-nsf-code?
"Given an error code, return truthy if it is NSF."
[code]
(#{"BE1" "BE2"} code))
Clearly the result here could be nil or a string but it's definitely meant to
be used as a predicate. Similarly:
(defn nsf?
"Given the result of an SBW sale, return true if it failed with NSF."
[result]
(and (= "failure" (:result result))
(some is-nsf-code? (:errors result))))
Again, the result could be false or nil or a string but is meant to be used as
a predicate.
As an aside, for core.typed, we annotate the first as [String -> Boolean] with
^:no-check so it type checks as a true/false predicate and then we annotate the
second as [SBWResult -> (Nilable Boolean)] and that's all fine... but is it
"good style"?
[1] http://dev.clojure.org/display/community/Library+Coding+Standards
[2] https://github.com/bbatsov/clojure-style-guide#naming
Sean Corfield -- (904) 302-SEAN
An Architect's View -- http://corfield.org/
"Perfection is the enemy of the good."
-- Gustave Flaubert, French realist novelist (1821-1880)
signature.asc
Description: Message signed with OpenPGP using GPGMail
