This page is part of the web mail archives of SRFI 50 from before July 7th, 2015. The new archives for SRFI 50 contain all messages, not just those from before July 7th, 2015.
> From: bear <bear@xxxxxxxxx>
> On Sun, 25 Jan 2004, Tom Lord wrote:
> > > However, in either case you still have two problems:
> > > 1) Both are very different from unconditional O(1) access.
> >The language I recommend for R6RS says "expected case O(1)", not
> >unconditional.
> >It's not a requirement, just guidance -- so it doesn't prevent any
> >implementation from conforming.
>
> If it's guidance rather than a requirement, it would be better
> to use the word "recommended" rather than "expected". The latter
> has a technical meaning when talking about algorithmic complexity,
> which is the expected runtime of an algorithm for "normal" data.
?
I'm using "expected" in the technical sense you refer to.
The proposed guidance is about expected, not worst-case performance.
(At the same time -- this is turning silly. It would be
unprecedented to have performance-related guidance in R^nRS
so perhaps the simplest thing is just to let the precedent stand.)
Just consider me as having walked around in the commons for a while
carrying a big sign that says "Don't make lame implementations of
Scheme strings."
-t