[sc34wg3] One standard or several?

[Mason, James David (MXM)]

I've already declared myself to be of the one standard, multiple parts
school. If it defines Topic Maps, it ought to be one standard. If something
has to conform to it (whether for interchange syntax or interpretation of
what do do when one gets something in that syntax), it needs to be one
standard.

Lars Marius offers several options below. Like him, I think some of them
(particularly "a" and "b") are pretty bad. His "c" takes us back to where we
were in 2000. His "d" is where I thought we were headed until Charles
started pushing the multiple standards bit. It's what we drafted the NP for,
though we eventually checked a different box about what the results should
be. I agree that "e" would be a mess. I don't think we have an option to do
"f" (it's possible to turn a failed DIS into a TR, as WG2 did with Font
Services, but not to turn a published IS into one).

If what Lars Marius says in one of his most recent messages about the
relationship of the current syntaxes to SAM and thence to RM is accepted, I
think we can get to "d" in steps. That is to say, we can publish the SAM and
RM, then revise the current 13250 and split it into syntax, tutorial,
conformance, and whatever else. Until then, the current 13250 can stay where
it is, and people can conform to it as best they can.

I really don't like multiple standards, particularly if there are strong
dependencies among them. It's one thing to say, for example, that the
interchange syntax for some data structure makes normative reference to
UNICODE, a separate standard, and an entirely different sort of thing to say
that the semantics of an interchange syntax are in a standard with a
different number.

Jim


-----Original Message-----
From: Lars Marius Garshol [mailto:larsga@garshol.priv.no]
Sent: Wednesday, January 22, 2003 10:32 AM
To: sc34wg3@isotopicmaps.org
Subject: Re: [sc34wg3] One standard or several?

<snip>

I liked the multipart approach because it seemed clean and easy, and
because the components N323 proposes to put in a single standard all
seem (to me) to belong in the core standard.  It's not a big deal for
me, however.

The problem I see with multiple standards is what is going to become
of ISO 13250. These are the possibilities I can think of:

 a) Make the RM be 13250. This is likely to confuse people, as the
    current 13250 and the RM are only distant relations of one
    another, and other parts of the family are much closer to what
    used to be 13250.

 b) Make the SAM be 13250. This has the same problem, although less
    so.

 c) Make the HyTM syntax specification be 13250. This would make 13250
    continue to specify the same thing, but it would no longer stand
    on its own.

 d) Withdraw 13250, as it will be replaced by new and better
    specifications.

 e) Keep 13250 pretty much as it is. This means turning what is
    essentially a tutorial into a normative text with hairy relations
    to the other standard. I think this option is technically and
    editorially a horrible option, and find it very difficult to live
    with. 

 f) Keep 13250 pretty much as it is, but turn it into a technical
    report that is basically a technical topic map tutorial, pretty
    much like XTM 1.0, but with more explanations.

Personally, I don't like any of these options, but I do think f) is
the best choice, provided ISO procedure actually allows it. I think a
multipart standard is much cleaner, and I don't think there is any
difference in marketing between the two approaches.