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

From: Eli Barzilay (eli at barzilay.org)
Date: Fri Feb 18 09:13:45 EST 2011

Three minutes ago, Matthias Felleisen wrote:
> Ryan, thank you very much for refocusing the discussion about
> testing on the essence and away from the superficial syntactic
> issues.

These are two completely unrelated issues, very concretely: one is the
representation layer that would allow different tools to cooperate,
the other is what the user of a particular test suite uses.

Neither of them is superficial.

The "eli-tester feedback" subject is unfitting for this discussion.

> Syntax, as we all know, is just a macro [which some people may get
> backwards and conclude that testing is just a macro. Nothing could
> be further from the truth, however].


The syntax is important, and this was never said seriously.  Now that
the two subjects are clear (hopefully), the original point was more
like this:

* You need some representation for data related to testing.

* You need some syntax (more generically, some api) for writing tests.

The difficulty of the first is in designing this representation, the
implementation is trivial (it'll end up as a bunch of structs and some
parameter/s).  The difficulty of the second is in both design (for
example, making tests catch too many errors), and syntax (making them

In both cases, if you approach the design in a naive way, then testing
becomes that trivial little bit of code, and some api that is easy to
do.  But since the design is not as straightforward, this is wrong,
which is why it was never serious.  (Otherwise I'd never keep my thing
into a hidden unpublic tool.)

So can we finally get over this "just a syntax" joke now?

          ((lambda (x) (x x)) (lambda (x) (x x)))          Eli Barzilay:
                    http://barzilay.org/                   Maze is Life!

Posted on the dev mailing list.