type disjointness Alex Shinn (02 Jan 2019 06:47 UTC)
Re: type disjointness John Cowan (02 Jan 2019 13:27 UTC)
Re: type disjointness Per Bothner (06 Jan 2019 07:10 UTC)
Re: type disjointness John Cowan (07 Jan 2019 14:21 UTC)
Re: type disjointness Arthur A. Gleckler (07 Jan 2019 16:13 UTC)
Re: type disjointness John Cowan (07 Jan 2019 20:59 UTC)
Re: type disjointness Per Bothner (10 Jan 2019 20:19 UTC)
Re: type disjointness John Cowan (10 Jan 2019 22:38 UTC)
Re: type disjointness Per Bothner (11 Jan 2019 01:04 UTC)
Re: type disjointness John Cowan (11 Jan 2019 01:50 UTC)
Re: type disjointness Per Bothner (11 Jan 2019 19:41 UTC)

Re: type disjointness Per Bothner 06 Jan 2019 07:09 UTC

I think it would be helpful to allow, but not require, that
the uniform vector types be vectors.  Specifically, to allow that
both (TAGvector? VAL) and (vector? VAL) may be true for the same VAL,
but that (TAG1vector? VAL) and (TAG2vector? VAL) may not both be
true for different TAG1 and TAG2 (except bytevector and u8vector).

Consider a procedure that works on array indexes, such as build-array in srfi 164.
On some systems (including Kawa) it would be desirable to use s32vector
for the indexes, and to use use s32vector-ref to access a specific index.
But that reduces portability.  It would be beneficial to be able to use vector-ref,
even if the argument is a s32vector.  (Of course, using vector-ref
may be less efficient than using s32vector-ref.)
--
	--Per Bothner
xxxxxx@bothner.com   http://per.bothner.com/