On Tue, Aug 29, 2000 at 07:34:10PM -0700, Nathan Wiger wrote:
> Mark-Jason Dominus wrote:
> > 
> > The IMPLEMENTATION section of the RFC is supposed to be mandatory, but
> > there have been an awful lot of RFCs posted that have missing or
> > evasive IMPLEMENTATION sections.
> 
> Well, I have to counter this with the following: Having a complete
> IMPLEMENTATION section should wait until at least v3. 

Then there are the "position paper" RFCs that don't require an 
implementation.  Simon's RFC 28: 'Perl should stay Perl' is a perfect example.

And don't forget the "really cool new ideas" where a discussion of 
implementation is premature, like Damian's RFC 21: 
'Replace C<wantarray> with a generic C<want> function'.

Both of these should have a stub, but should never be expected
to have a full implementation, v3 or otherwise.

> So I would say, v1's and v2's should get some slack. If the idea's still
> around by v3, then IMPLEMENTATION should be considered. But doing this
> before them is a waste of time, IMO.

Yes, there should be some slack, but not on the reqirement to have an
implementation section.  

I would hope from here on out, RFC authors spend a modicum of time
writing up why an idea doesn't need an implemenation, why such
discussions are premature, or why the author needs help writing it up.

Any requirements on how solid an implementation section should be
should be left to the working group chairs.

Z.

Reply via email to