Can't you just assert on the return value of spec.test.check in a clojure.test test? ie, this is a test in one of our repos: (deftest ^:spec generative (is (not-any? :failure (st/check [`export/invert-table->account-ids])))) On Friday, September 29, 2017 at 3:36:26 AM UTC-7, Khalid Jebbari wrote: > > I've managed to get something working. Very dirty and hardcoded most > things though, so not reusable at all. The key here is to parse the return > of spec/describe and retrieve the value for the :arg, convert it to real > spec with eval and retrieve its generator. > > (defspec a-test > (let [args (nth (s/describe 'foo.core/bar) 2) > spec-code (map #(if (symbol? %) (symbol > "clojure.spec.alpha" (str %)) %) args) > spec (eval spec-code) > args-gen (s/gen spec)] > (prop/for-all [argz args-gen] > (s/valid? ::ret-spec (apply foo.core/bar argz))))) > > > In the end one shoot write a proper spec/describe parser ? > > Or has someone a better idea ? > > > On Friday, September 29, 2017 at 12:08:57 PM UTC+2, Khalid Jebbari wrote: >> >> Hello, >> >> I'm struggling to find a way to to use the fdef specs I wrote in >> clojure.test tests. I can run them fine in the repl with spec/exercise-fn >> or spec.test/check, really nice when developping by the way. Now that I'm >> happy with the result I'd like to encode this knowledge in tests to prevent >> regressions. I don't need more tests that this, not specific property etc. >> >> I found no way to plug the spec.test/check in clojure.test or easily >> reuse fdef specs. test.check/defspec and quickcheck expect properties as >> their argument. spec/describe return a LazySeq that I found hard to exploit >> without a lot of manual wiring, parsing and trial-and-errors. >> >> If I had to write it by hand, it would look like : >> >> (defspec myspec 100 (prop/for-all [one (spec/gen ::first-arg) >> two (spec/gen >> ::second-arg)] >> (is (true? (spec/valid? >> ::ret-spec (myfunc one two)))) >> >> The problem is that it's incomplete with regards to spec possibilities : >> spec/or, spec/nilable etc. and I use them. Also I the function changes (in >> any way) the test becomes irrelevant instantly. >> >> A colleague resorted to manually calling spec.test/check in clojure.test >> and manually verifying the output of the function (the :result boolean, the >> :num-tests etc.). Feels way too manual, and doesn't report the shrunk value >> as nicely as test.check does. >> >> >> Maybe I missed something completely. spec/describe seems the best bet to >> introspect the spec and use it in for-all calls. But still too manual. >> >> Any help much appreciated. >> >
-- You received this message because you are subscribed to the Google Groups "Clojure" group. To post to this group, send email to clojure@googlegroups.com Note that posts from new members are moderated - please be patient with your first post. To unsubscribe from this group, send email to clojure+unsubscr...@googlegroups.com For more options, visit this group at http://groups.google.com/group/clojure?hl=en --- You received this message because you are subscribed to the Google Groups "Clojure" group. To unsubscribe from this group and stop receiving emails from it, send an email to clojure+unsubscr...@googlegroups.com. For more options, visit https://groups.google.com/d/optout.
