Re: ASCL a doomed idea?

[Jeffrey Carter <jrcarter@acm.org>]

Ted Dennison wrote:
> 
> Anyway, JC01 presents an unsafe style iterator. His comments say it could go
> either way, but if its safe, there's no sense in supplying the list with it like
> all JC01's routines do, so I think his presentation only makes sense for an
> unsafe iterator. For an unsafe iterator, I think "Position" is reasonable, as it
> at least underlines that you can't expect the compiler to do all your checking
> for you like you would with a true iterator.

It presents a specification that doesn't bias the implementation. If you
leave the list parameters off, you can't have an unsafe implementation,
but putting them in doesn't prevent a safe implementation.

> Which one we use is something that still needs to be decided. Perhaps there
> hasn't been much discussion about this because everyone is happy with the way
> the latest Strawman does it (safe)?

I'm not too concerned either way. Complete safety means insertion and
deletion are O(M) in time, where M is the number of Locations in
existence. The additional protection this gets you over mostly safe
Locations seems to be against pathological cases, misuse of a list of a
kind that I've never actually seen in practice (and I've seen some
pretty poor code using lists with completely unsafe Locations). So, I'm
not sure complete safety is worth the effort, but I won't be bothered if
it's deemed essential for the standard list.

-- 
Jeff Carter
"My brain hurts!"
Monty Python's Flying Circus