Giters

tgutu / clunit

CLUnit: A Common Lisp Unit Testing Framework.

Home Page:http://tgutu.github.com/clunit

Geek Repo:Geek Repo

Github PK Tool:Github PK Tool

wishlist: a version of assert-equality with a default argument

tpapp opened this issue · comments

commented

It would be more convenient to write

(deftest foo (bar)
  (assert-equality #'my-fancy-equality value1 expression1)
  (assert-equality #'my-fancy-equality value2 expression2)
  (assert-equality #'my-fancy-equality value3 expression3))

as something like

(deffixture bar (@body)
  (let ((*assert-same-test* #'my-fancy-equality))
    @body))

(deftest foo (bar)
  (assert-same value1 expression1)
  (assert-same value2 expression2)
  (assert-same value3 expression3))

The wrapper could also be something like

(deffixture bar (@body)
  (with-assert-same-test (#'my-fancy-equality)
    @body))

Obviously the function/macro/variable names are just examples and it is very likely that you can come up with something better. This functionality is present in other libraries, eg lift:*lift-equality-test*.

commented

I will try and get to this by the end of the month. Been bogged down with other things.

commented

I decided to go with the first method clunit:*clunit-equality-test*. The update to the ASSERT-EQUALITY macro is unfortunately not backwards compatible. See http://tgutu.github.com/clunit/api/clunit.xml#assert-equality

I have also marked the other ASSERT-EQ* macros as depreciated in the API. I think going forward, ASSERT-EQUALITY should suffice.

N.B. I have updated the documentation on how to use the test dependency option by adding an example http://tgutu.github.com/clunit/#clunit_12

commented

Thanks for the change. But I would not insist on deprecating the other assert-eq... forms: I like the fact that they don't use a global variable and thus they lead to very clear code when a CL equality operator suffices for testing. To tell the truth, I also liked assert-equality in its previous form; it just got tedious when I was comparing many things with the same operator. But for a single use, wrapping it in a let is even worse.

All in all I would prefer if you reverted this change, and optionally re-introduced the new assert-equality under a different name (maybe assert-equality*), making the old one available under its old name. This would also lead to a backward-compatible change.

Sorry for not communicating my suggestion more clearly.

commented

As soon as you said that, it made so much more sense. Made the change immediately before anyone pulled I hope.