On Tue, 20 Sep 2005, Michael Sperber wrote:

> Yes, but you misunderstand the sentence in the draft: It's not saying
>
> "If ..., it would still be possible to <something>.  However, we want
> to make <something> impossible."

If I misunderstood it, it was not in this way :-)

> Instead, the sentence is trying to say: "If the construction procedure
> was defined outside the DEFINE-TYPE, it wouldn't be available through
> the type defined by DEFINE-TYPE, and we couldn't use it, say, through
> the PARENT clause."
>
> Would this be a better wording?

It would be better, although strictly speaking I'm not sure you should say
that it is available through the type, since that can be misunderstood as
implying that it is in the type descriptor, which it is not.

The only reason for the rationale is the formals in the parent clause.  My
statements were in the context of a suggestion to drop the idea of
initialization via constructor formals, maybe instead just using labels as in
SRFI-9.  That would have taken away the reason for the rationale.

Cheers
Andre