[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: Procedural vs. inspection, lists vs. vectors (2 of 3)

It is time to resolve this SRFI one way or another.

John Cowan wrote:
> Given that an implementation of SRFI-99 can't really change its
> implementation in any interesting way if one uses only the procedureal
> and/or syntactic libraries but not the inspection one, I propose folding
> the inspection library into the procedural one.  In order to support
> rtd-{accessor,mutator} with non-constant arguments, one must have access
> to the rtd object at runtime anyway, including its field list.  I think
> the separation is an unnecessary carry-over from R6RS.

I think you're right.  The structure of this SRFI was
deliberately patterned after the structure of the R6RS
system.  I think that makes the two systems a little
easier to compare, which was one of this SRFI's purposes.

> I add my voice to the call for using lists rather than vectors in the API
> for make-rtd, rtd-constructor, rtd-field-names, and rtd-all-field-names,
> of course without prejudice to the use of vectors or vector-like objects
> internally.

I think the use of vectors is another carry-over from
the R6RS.  I don't care about it one way or another.
If the SRFI editor(s) in charge of this SRFI suggest
I make that change, I will.