Re: minutes from ER TC teleconference 20010516

[Norman Walsh <ndw@nwalsh.com>]

/ Lauren Wood <lauren@softquad.com> was heard to say:
| John raises an issue: why we treat system IDs and other URIs separately.

I went looking for discussion of this issue (I know we had some) but
there's no followup to the most relevant message[1] that I find. I
think the issue got swallowed by our discussion of what to do with W3C
XML Schema.

Here's my answer to John's question.

We treat them differently because system identifiers and URIs are
different.  While it's true that all system identifiers must be URIs,
the converse is not true. More to the point, the XML REC speaks of
public and system identifiers, TR9401 catalogs speak of public and
system identifiers, and users speak of public and system identifiers.

Despite the fact that one can make the argument "but all system
identifiers are URIs, so let's just treat them as URIs" I think that
it would be confusing to users to do so.

I am also very happy with our current clean split between external
identifier lookup (which carefully preserves TR9401 semantics in an
understandable way) and general URI lookup.

Yes, URIs and system identifiers could be conflated. But I think it
would confuse users and I can see no benefit in doing so.

In considering the use cases, I can see only one case in which it
would make any practical difference. If you had a URI that you wanted
to treat as *both* a system identifier and a generic URI, you'd have
to have two entries for it. But I don't think that happens in
practice. And if it did, might you not want to map it differently?

                                        Be seeing you,
                                          norm

[1] http://lists.oasis-open.org/archives/entity-resolution/200102/msg00040.html

-- 
Norman.Walsh@Sun.COM   | As a general rule, the most successful man in
XML Standards Engineer | life is the man who has the best
Technology Dev. Group  | information.--Benjamin Disraeli
Sun Microsystems, Inc. |