[ProgSoc] Progsoc debate?

Nathan de Vries nathan at atnan.com
Wed Apr 16 10:11:55 EST 2008


On Wed, 2008-04-16 at 09:15 +1000, Myles Byrne wrote:
> That's good except I don't really want to focus on types
> (loose/strict/explicit/inferred or even duck).

Good point. I guess my reasoning for avoiding dynamism was that any
decent Java developer can do some pretty flexible things with the
reflection API, and runtime casting / type tests are inherently dynamic.
Their static (for the most part) type system makes code analysis &
modification tools possible.

That being said, as we speak there are a million and one postgrad
students labouring away on theses offering dramatically different
categorisations of type systems, so you're probably right in avoiding
types.

> The often touted benefits of dynamic (think runtime, not types)
> languages are negated by the use of powerful code analysis and
> modification tools that come with modern static (hard to modify
> runtime) languages.

Sounds good.

> ...the advantage of the language *not* changing is that you can
> actually write the tools to generate the getters/setters and that
> functionality wont be pulled out from under you with the next language
> release (which is years out anyway). The language essentially acts
> like a high level assembly and tools (not text) are what you use to
> interface with your program.

This might be a bit of a misnomer, because while Java is (sometimes
arguably) backwards compatible at the bytecode level, language features
that the tools need to refactor are not. For instance, if you want to
start refactoring some code that uses generics, your tools will need to
understand what generics are (i.e JDK5.0+) despite the fact that the
compiler is turning your generics code into the same bytecode that you
would have been producing without generics support.

> I heard on a podcast the first version was written in smalltalk, then
> they gradually swapped bits out for Java implementations.

If you can find out what podcast it was, I wouldn't mind having a
listen.


Cheers,

--
Nathan de Vries
-------------- next part --------------
A non-text attachment was scrubbed...
Name: smime.p7s
Type: application/x-pkcs7-signature
Size: 1878 bytes
Desc: not available
Url : http://progsoc.org/pipermail/progsoc/attachments/20080416/81fdd1e3/attachment.bin 


More information about the Progsoc mailing list