Hi Steph

On Tue, 22 Jan 2002, ducasse wrote:

[snip]
> > Perhaps the right, if slow, solution is to generalize the mechanism so
> > that #asFoo expands/is DNU trapped to Foo fromReceiverType: receiver.
> 
> NO NO NO.
> 
> This is not we have DNU that we should use it for basic engineering
> problems.

Er...notice that there were at least two solutions :) Neither is, IMHO,
workable. But I think I prefer either than forcing myself to use
inconvenient messages (if #as: really does turn out to be
inconvenient). After all, DNU trapping is merely an implementation
technique, in this case. It can be refactored easily.

> DNU is a meta approach that will hamper code analysis and make
> things to complex.

I like to avoid that when I can. But I'm not a code analysis guy, so the
pressure on me from that front isn't so great :)

I will say that I've long wanted some selector matchers that looked inside
symbols (both in selector names and in arguments), but Ah Well :)

> This is also why this is cool to use changeClass but why we should not using
> while implementing state pattern.

Hmm. I didn't catch this reference. Details?

Are you agains the general idea (surface syntax of #asFoo, transformed to
as: foo?) or just a DNU implementation?

Cheers,
Bijan Parsia.