On Sat, 05 Sep 2009 21:34:44 -0400, John Cowan <[email protected]> wrote:
> Aaron W. Hsu scripsit:
>
>> I agree that we should definitely standardize the things that are ripe.
>> I
>> also think that we shouldn't start trying to throw every standard into a
>> single document or process. It makes sense to me to standardize all
>> these
>> things in parallel, as separate efforts.
>
> That's been *done*. That's what the SRFI process is. Now I'm talking
> about folding back a small number of SRFIs and a handful of stuff from
> R6RS, so that people can look at a single document and see all the
> basics.
>
> As I've said, I'm okay with making lots of things optional: R5RS already
> does, far beyond just the six procedures actually marked "optional",
> and the charter suggests even more. But I am not up for fragmenting just
> for the sake of fragmenting, so that someone who supports only "lambda"
> can claim he has a Scheme, nyuk nyuk nyuk.
I'd be okay if we let R6RS Base or R5RS be the basic starting point. I
don't know if everyone else would, but I think it could be made to happen.
I also agree that the SRFIs should be the first candidates for
standardization. But trying to add things to the Core or lumping more
changes into a single standard makes it that much more difficult to get it
standardized. Lets just standardize those important SRFIs and be done with
it, with or without the rest of the standards.
>> So, we could standardize records outside of the core Scheme. Doing so
>> would get us the benefits of a standard record type, but would simplify
>> the process of standardizing a core.
>
> Only in the sense that you simplify the job of peeling potatoes by
> dividing them into five lots first. If you *really* have parallel
> processors available, that's fine; but history suggests that we simply
> don't.
It makes it easier to get things approved. There is a psychological
barrier to making this big standard and then waiting a while to get full
approval. It's easier to get them done in small chunks. At least, though,
I agree that optionality is good.
>> So these are things we definitely want to standardize. Let's take
>> advantage of the very modular aspects of working groups and standardize
>> these features outside of the core as libraries, and restrict the core
>> to a fairly clean, lean semantic model and some other more academic
>> things.
>
> What things? The line must be drawn somewhere. The SC has drawn it
> at IEEE; I'm trying to leverage the SRFI and R6RS processes to get the
> good stuff in. Why restandardize records, or string ports? We *have*
> standards for them.
I agree. Let us get them through, and let the other things fight their own
battles.
Aaron W. Hsu
--
Of all tyrannies, a tyranny sincerely exercised for the good of its
victims may be the most oppressive. -- C. S. Lewis
_______________________________________________
r6rs-discuss mailing list
[email protected]
http://lists.r6rs.org/cgi-bin/mailman/listinfo/r6rs-discuss
