On Tuesday 20 February 2001 16:51, Dan Sugalski wrote:
> Honestly, the PDDs are for the stuff that was implemented, not the stuff
> that was decided. Or, more clearly, PDDs describe the implementation or
> proposed implementation at the internals level. RFCs are for
language-level
> features.
It should sort of do both. Okay, maybe arbitrary language decisions
needn't be P?Dd, but there should be some documented consensus, even for a
language feature, of what is (or will be), if only to give the developers a
clear target to shoot for.
>
> We should have PDDs on garbage collection and memory allocation (I know,
I
> know--I'm working on it! :). We should not have PDDs on, say, currying.
Well, a well-designed spec on currying language would certainly help with
development of the internals to handle currying. It would also provide a
nice starting point for the currying documentation.
Lest the architect say, "Build me a house", and then complain that we don't
match the plans he never gave us.
--
Bryan C. Warnock
[EMAIL PROTECTED]