[racket] tests/eli-tester feedback (Was: Racket unit testing)

On Tue, Feb 15, 2011 at 8:35 AM, Eli Barzilay <eli at barzilay.org> wrote:
> 5 minutes ago, David Van Horn wrote:
>> On 2/15/11 7:02 AM, Eli Barzilay wrote:
>> > Three hours ago, Hendrik Boom wrote:
>> >> On Mon, Feb 14, 2011 at 04:43:24PM -0600, Robby Findler wrote:
>> >>> I guess I'm asking if true is bound somewhere or not.
>> >>
>> >> It almost looks as if you want predicates on the right side of =>'
>> >
>> > Yes -- the whole point of the simple arrow-less version is to make
>> > using predicates very easy.  For example, instead of some
>> >
>> > 1. (test E1 =satisfies>  even?
>> >           E2 =satisfies>  (lambda (x) (>  x 50)))
>> >
>> > you'd write
>> >
>> > 2. (test (even? E1)
>> >           (>  E2 50))
>>
>> You could allow the arrow-less form (which I like) but require it to
>> produce #t and only #t.  This accommodates your examples while catching
>> things like:
>>
>>     (test (fact 5) 121)
>
> Nice idea -- and for the rare cases where you want a non-boolean
> predicate you'd resort to the less conveniet
>
>  (test (and (member E '(1 2 3)) #t))
>
> Robby: would you find this acceptable, or is the hole that it leaves
> (making the no-arrow mistake with a #t-resulting expression) too big?

I wouldn't have suggested it :), but it seems like a much more minor
issue than it was before.

The main thing I'd like to see at this point is docs and some decisions made!

Robby