16:33 <ashpool> I have a very simple s-exp-based DSL implemented in Racket. What's the best way to write a test-suite for it using rackunit? I have customized #%module-begin, and I want to test the values it returns.

16:34 <ashpool> I've studied namespaces and sandboxed evaluation for a few days, and I've kind of made it work, but it's very ugly and I have a feeling in must be much much simpler than that.

16:35 <ashpool> I would love to look at some code samples for what a test suite might look like, if available.

16:51 <ashpool> Here's a simple example. https://paste.sr.ht/~ashpool/a94be43beafb0baa28a27c58113ea6008ea3a9ad

16:52 <ashpool> I'm dynamic-requiring a module, and when I do that, a value is printed to the standard output. How do I get hold of that value in "rade-test.rkt" so I can write tests against it? (dynamic-require) itself returns void.

16:54 <bremner> hmm. Can with-output-to-string help?

16:55 <bremner> I don't really know anything about dynamic-require.

16:59 <ashpool> I could modify the language so that it exports a value, say 'result, and then do (dynamic-require '(submod "rade-test.rkt" zzz) 'result). But what if I don't want to export anything? And what if I need to do the same for a non-s-exp-based language?

17:00 <bremner> I basically only use (module+ test ...)

17:04 <ashpool> Yeah that's a good idiom but unfortunately my language is so minimalistic it doesn't provide things like (module... ). So I need to write a separate module in racket, make it have a test submodule, and in that submodule dynamically require and test the program in DSL. At least that's how I see it.

20:38 <dzoe> ashpool: why not plain old require?

20:40 <dzoe> (module my-mod racket/base 1)(require 'my-mod) correctly outputs 1 in REPL

20:40 <dzoe> I don't see why it should be different for any module. As long as the whole module evaluates to some final value (its last expression).

20:42 <dzoe> Ok, for capturing the value, dynamic-require does the trick (for external modules as well).