I just realised that if you're using a tools.namespace based workflow, the spec registry isn't cleared when you refresh. Is there a recommended way to "clean the registry" safely?

I suspect that if I just reset! the registry I might be missing internal/clojure.core.specs specs?

Question. This will break with Clojure telling me that “No such var: eid”

(s/def ::eid (s/with-gen (s/and number?
                                pos?
                                #(instance? java.lang.Long %)
                                (complement nil?))
               gen/int))

(s/fdef foo/bar
        :args (s/cat :a any?
                     :b ::eid)
        :ret (s/coll-of ::some/things))

what is es/eid?

::eid is the first form

… sorry wrong namespace

fixed

Why can’t I use ::eid in my function spec :args?

@twashing I can load this into a REPL fine:

(defn bar [])

(s/def ::eid (s/with-gen (s/and number?
                                pos?
                                #(instance? java.lang.Long %)
                                (complement nil?))
                         gen/int))

(s/fdef bar
        :args (s/cat :a any?
                     :b ::eid)
        )

It’s the function that generates generative tests, based on the function spec.

... one sec, lemme find it.

clojure.spec.test/check

::stuff is a map with a fixed key :data that will never change,.

(s/def :foo/data (s/keys :req [:relevant/key :another.relevant/key]))
(s/def ::stuff (s/keys :req-un [:foo/data]))

what’s the recommended way to run spec checks during automated testing? we’re currently using lein test with clojure.test and it would be nice to use something like the solution here https://stackoverflow.com/questions/40697841/howto-include-clojure-specd-functions-in-a-test-suite

if I have to I can copy & paste and adapt that code but it would be nice to have a supported library

@joost-diepenmaat if you paste it into https://github.com/gfredericks/schpec and make a PR, I'll release it not sure if that counts as "supported" or not

hi everyone! I'm trying to understand how to match :clojure.spec.alpha/problems inside :clojure.spec.alpha/value in explain-data result independent of the specs that were applied and I'm having some trouble

so the thing is that seeing a problem in [::k 1] you don't know if it means, it's the value of that key, or it's the element at pos 1 of a seq in the value of that key

does it make sense?

see https://dev.clojure.org/jira/browse/CLJ-2192

you're right in your analysis, it's been discussed before and it's logged in that ticket

great, thx a lot @bronsa

So I'm wondering, is spec with s/def + s/valid? supposed to replace predicate functions? Otherwise I find it hard to decide what should be a spec and what should just be a function

Oh, I suppose that isn't really a good question. More pertinent would be whether new predicates should be written as anon functions in s/def or should s/def refer to a named predicate function

I am wondering how I would write a spec for a nested map. I have something like this {:1 {:user-id 1 :username "bob"} :2 {:user-id 2 :username "jim"}} Where I have a (s/def ::user (s/keys :req-un [::user-id ::username])). I know how to do this if I had a vector of users by using (s/coll-of ::user), but how would I approach this with having the map keys being the ::user-id spec and the values of each map being the ::user?

@assoc-in (s/map-of keyword? ::user)

@bfabry Ah yes I didn't see that looks like just what I needed. Thanks!

np