[R6RS] Re: Why there are two syntactic layers

Michael Sperber sperber
Thu Jun 23 03:38:52 EDT 2005

"Manuel Serrano" <Manuel.Serrano at sophia.inria.fr> writes:

> I understand this but what I don't understand is why you don't like the
> Marc's proposal. As far as I understand it, it supports both a simplicity
> (by using implicit defaults) and fanciness. Am I wrong?

It's not about the possible simplicity of the uses of the record
proposal, it's about the complexity and redundancy in the record
proposal itself.  (Also, looking at a DEFINE-TYPE form, how do I
determine whether it's "simple" or "fancy"?---You and I probably have
very different ideas.)

Check out just the sheer size of SRFI 9, both the description and the
reference implementation.  Compare with the size of the DEFINE-TYPE
description and its implementation.  Clearly, DEFINE-TYPE is much
larger than it needs to be in order to fulfill the requirements we set
forth in Boston.  That makes it unsuitable for the core.  In the
library, we can go crazy, if need be.  (Even though, given the limited
experience we have with fancy record forms, we should probably leave
some of that for R7RS, with some SRFIs to feed it.)

I'll admit that even DEFINE-SIMPLE-TYPE is more complex that it needs
to be---that's an artefact of the "smooth upgrade" constraint and the
resulting syntax.  But it's still much simpler than the fancy version,
andat least it has no redundancies.

So---I personally would be happy to change the syntax as long as it
meets the constraints I mentioned.  I'm also not averse to adding
stuff to the fancy layer.  (Kent probably has more to say on the
matter.) Moreover, we can have a discussion over whether the "smooth
upgrade" approach is so important.

However, over the question of simplicity, redundancy, and layering, I
don't see an alternative approach that's likely to get us consensus.
Maybe this approach isn't going to get us consensus either, but we
sure are completely deadlocked over the "one true record syntax"
issue: The records I want are unacceptable to you, and the records you
want are unacceptable to me.  Do we still want to make progress?  I

Cheers =8-} Mike
Friede, V?lkerverst?ndigung und ?berhaupt blabla

More information about the R6RS mailing list