Missing "tag" in EBNF? Wolfgang Corcoran-Mathe (11 Apr 2026 16:19 UTC)
Re: Missing "tag" in EBNF? John Cowan (11 Apr 2026 18:45 UTC)
Re: Missing "tag" in EBNF? Arthur A. Gleckler (11 Apr 2026 19:00 UTC)
Re: Missing "tag" in EBNF? John Cowan (11 Apr 2026 19:47 UTC)
Re: Missing "tag" in EBNF? Peter McGoron (11 Apr 2026 19:52 UTC)
Re: Missing "tag" in EBNF? Wolfgang Corcoran-Mathe (11 Apr 2026 19:52 UTC)
Re: Missing "tag" in EBNF? Wolfgang Corcoran-Mathe (11 Apr 2026 20:02 UTC)
Re: Missing "tag" in EBNF? John Cowan (11 Apr 2026 20:30 UTC)
Re: Missing "tag" in EBNF? Wolfgang Corcoran-Mathe (11 Apr 2026 21:45 UTC)
Re: Missing "tag" in EBNF? John Cowan (11 Apr 2026 23:06 UTC)

Re: Missing "tag" in EBNF? John Cowan 11 Apr 2026 20:30 UTC

On Sat, Apr 11, 2026 at 4:02 PM Wolfgang Corcoran-Mathe
<xxxxxx@sigwinch.xyz> wrote:

> But how is 'read-array' to map an unknown tag to a storage class?

I think that is an "it is an error" case.

> I'm afraid that, if the tag spec is made inclusive, every array with
> a tag that doesn't correspond to one of the SRFI 231 storage classes
> will end up getting 'generic-storage-class'--which is certainly not
> what the syntax would lead you to expect.

If you didn't expect that, what would you expect from "#ai22" or
"#aq64" or "#anoodle"?  I think those are the three types of
possibilities: unknown element size, unknown element type, or just
plain unknown.  One implementation could throw an exception; another
could widen the size to the next larger size and the element type to
generic; a third could widen all unknown tags to generic.