Squeak-dev Digest, Vol 22, Issue 20

Andreas Raab andreas.raab at gmx.de
Sun Oct 17 22:35:56 UTC 2004


Andres,


> AR> I respect your opinions but please don't put things into my mouth that 
> I
> AR> haven't said or meant.
>
> I did not put words in your mouth.  I expressed what I understood you
> meant.  I'd like to know how I didn't get what you wrote.

Well, for one thing, I haven't even said what I expect from a type system. 
For another, I haven't said whether it would be part of the language 
semantics or not (thanks to Stef for making this aspect clear to me - it was 
clear to me that a type system should not be used to do stupid things like 
early binding -which is what happens when you claim "performance gains"- but 
saying that it does not affect the language semantics is the clearest way of 
stating it). As such, it follows immediately that such a type system would 
be optional (if it doesn't affect the language semantics you might as well 
not use it) and that there may be various type systems at the same time. In 
other words I fully agree with Stef's term of an "optional pluggable type 
system".

>>> From another point of view: Smalltalk's point was to teach kids, and
>>> it seems to me it was important to make it late bound.  Therefore I
>>> don't expect kids to understand the far reaching consequences of the
>>> static type system you propose.
> AR> But you are aware that eToys do have a static type system, are
> AR> you? It seems as if kids don't have that many problems with static
> AR> type systems as you are claiming.
>
> There was no eToys back in the 70s.  I thought the point of Smalltalk
> was to let kids write Smalltalk code.

Well, let me put it like this: The single largest (and repeated) complaint I 
heard Alan make about Smalltalk-80 is that "no child has ever programmed in 
it". That doesn't say anything about type systems but it says something 
about the goals. For Smalltalk-80 the goal was not to enable children to 
program in it. For eToys, it is.

> While I am not familiar with the static type system in eToys, I am not
> seeing how a static type system in eToys would be similar at all with
> a static type system for Smalltalk code either.

If you read the first paragraph (optional type system) and if you look at 
how the type system in eToys presents itself (you don't type a single extra 
word) it takes very little imagination to see where this might be headed. 
StrongTalk is another worthwhile example to look at.

Cheers,
  - Andreas




More information about the Squeak-dev mailing list