Animorphic ST (Strongtalk) released!

David Farber dfarber at numenor.com
Thu Jul 25 16:21:01 UTC 2002


Stephen -- Exactly. Ultimately, when I 'accecpt' a method I want Smalltalk to:

1) Parse and compile my code.
2) Infer types and typecheck the method.
3) Run Smalllint.
4) Look for "Design Smells".
5) Run all relevant unit tests.
6) Give me the results of all of the above in <250ms.

In other words, when I'm done with a method I want the computer to do all the leg work to figure out if I've missed something. I want to hear "I'm sorry Dave, I can't do that" if something's not kosher.

#6, of course, will be the tough one. Until I can get instant feedback, I'd be willing to have 2-5 as seperate tools that I run less frequently than after every method change.

david

At 09:31 AM 7/25/02 -0400, you wrote:
>To me, working with Strongtalk felt like it was a specialized form of
>SUnit testing.  For instance, one of the options in Strongtalk is to
>perform some form of type validation on a class (and I think there were
>several levels of checking available).  This to me felt like running an
>SUnit test suite.
>
>And, as I've said before, I think the typing should be optional for both
>the writer and the reader (meaning that any attempt to do this in Squeak
>should take into consideration that even though the author of a method
>may have included type annotations, the reader may not want to see
>them).
>
>My feeling is that the sweet spot for typing will fall somewhere between
>the Strongtalk system and SUnit testing.
>
>- Stephen
>
>
>
--
David Farber
dfarber at numenor.com



More information about the Squeak-dev mailing list