data/diff does not return a tuple when comparing different maps

Question

*Problem:* clojure.data/diff inconsistently returns a lazy seq when comparing different maps, but a vector otherwise.

user> (data/diff {:a 1 :b 2} {:a 1})
({:b 2} nil {:a 1})

This is inconsistent with doc and normal behavior :
user> (data/diff {:a 1 :b 2} {:a 1 :b 2})
[nil nil {:a 1, :b 2}]
user> (data/diff #{1 2 3} #{1 2 3})
[nil nil #{1 3 2}]
user> (data/diff #{1 2 3} #{1 2})
[#{3} nil #{1 2}]

The docstring states: "Recursively compares a and b, returning a tuple of [things-only-in-a things-only-in-b things-in-both]", implying that it should always return a vector.

Answer 1

Comment made by: yenda

Fixing it just requires to vectorize diff-associative output like this :

`
(defn- diff-associative
"Diff associative things a and b, comparing only keys in ks."
[a b ks]
(vec (reduce
(fn [diff1 diff2]

 (doall (map merge diff1 diff2)))

[nil nil nil]
(map

(partial diff-associative-key a b)
ks))))

`

Answer 2

Comment made by: alexmiller

There are other potential ways to address this, such as by using transducers instead. Not sure if that's worth doing, but seems reasonable to consider while we're making changes.

Answer 3

Comment made by: yenda

Maybe this could be done as an improvement and proposed in an other ticket.

Vec is already used to vectorize the lists in diff-sequential. I would suggest to just fix the bug and add the test cases that should have screen it.

Answer 4

_Comment made by: yenda_

There is a test case that should already fail :


[{:a #{2}} {:a #{4}} {:a #{3}}] {:a #{2 3}} {:a #{3 4}}


I get


({:a #{2}} {:a #{4}} {:a #{3}})

Answer 5

Comment made by: alexmiller

The test may need to be made more strict, checking not just for sequential equality but also for a returned vector.

Just curious - was this issue causing a problem in your code or did you just notice it and find it surprising?

Answer 6

Comment made by: yenda

Simple patch that just does for maps what is done for lists : Creates a new vector with the vec function.

Answer 7

Comment made by: yenda

@Alex Miller : I noticed a bug in my program behavior and traced it down to a (get diff 2) instead of (nth diff 2), but I realized that it was only buggy in some cases so I looked further and found out if was coming from diff.

Answer 8

Comment made by: yenda

More strict tests checking for a returned vector.

Answer 9

Reference: https://clojure.atlassian.net/browse/CLJ-1885 (reported by yenda)