Re: RFC - Prototype RFC Implementations - Seperating the men from the boys.

[Nathan Torkington]

Chaim Frenkel writes:
> We are not at that stage yet. 
> 
> There are too many new things that are _supposed_ to interact to
> bother with a prototype. It doesn't do any good, until the language
> is nailed down.

That's no argument against prototyping, though.  Prototype one feature
and then you can prototype any feature that uses it.  You have to
start somewhere, or you'll never start.

> You position is _perfectly_ and absolutely the only one that should be
> taken _ONCE_ the p6 code base has firmed up and is past the feature
> freeze phase.

Not necessarily.  Yes, we'll want to prototype things to answer
questions about their performance or semantics, but there's no reason
not to prototype things now.

Prototypes now have several benefits, in fact:
 - show what can be done
 - provide a basis to test and define interactions with other features,
   which is hard to do in the abstract
 - save us work later on

I'm not saying I think they should be mandatory, but an RFC with a
prototype behind it certainly carries more weight in my mind than an
unprototyped RFC.

Nat