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

Re: [Scheme-reports] Bytevectors should be called u8vectors

Aaron W. Hsu scripsit:

> > > SRFI 4: Racket, Gauche, Gambit, Chicken, Bigloo, Guile, Kawa, Scheme48,
> > > STklos, RScheme.  This information is probably out of date.
> > > 
> > > R6RS: Guile, Chez, Vicare, Larceny, Ypsilon, Mosh.
> >
> > As I suspected, common practice would favor the SRFI-4 API.
> I have to point out how I think this list is misrepresentative.  Racket
> supports the R6RS language as a built-in, so to have it on this list as
> an entry at all is misleading. It supports both, not one or the other.

As I already noted, I should have put Racket on both lists.

> This whole counting of implementations thing is a bit strange, and
> this list in itself is prettly close a count, and not of much help in
> this issue.

A list is not a count, for if it were, pairs would be numbers, and
we are explicitly told that they are disjoint.  :-)

The point of making a list is to provide facts about who implements
what.  Witnesses should be weighed and not counted, and I leave it up
to the reader how heavily to weigh any implementation.

> Outside of that, I think the name itself is a bikeshed 
> issue, and in R7RS we are not even talking about the larger issues of 
> data extraction from vectors of bytes. However, as I like the R6RS 
> mechanics (byte alignment rather than homogenous vectors) more than the
> other, I prefer that we use the name bytevector, so as not to confuse
> people in thinking that we are intending SRFI-4 style homogenous
> vectors. 

The term "blob" is neutral between the two semantics, and my BlobAPI
proposal provides both sets of operations on top of the single datatype.

> However, we have already had a long discussion of this in the lists. I
> am not sure that we are seeing anything new here. We made all of these 
> arguments before, and counter-arguments were also made.

I agree.

John Cowan   cowan@x  http://www.ccil.org/~cowan
Most languages are dramatically underdescribed, and at least one is
dramatically overdescribed.  Still other languages are simultaneously
overdescribed and underdescribed.  Welsh pertains to the third category.
        --Alan King

Scheme-reports mailing list