[sc34wg3] Backwards Compatability

Lars Marius Garshol sc34wg3@isotopicmaps.org
21 Oct 2001 18:19:06 +0200

The issue before us is not that of whether topic characteristics have
a single scope or a set of scopes, but that of whether the
foundational model for topic maps we are developing shall preserve
backwards compatibility with XTM 1.0 and ISO 13250.

I think we all agree that since the purpose of the foundational model
is to remove the ambiguities of those specifications, the removal of
some of these ambiguities will cause behaviour that before seemed
legal to now be illegal. This will be limited to minor issues of
processing, however, and not involve any changes to the XTM 1.0 model.

It seems to me self-evident that we cannot possibly fly the aircraft
and rebuild it at the same time. We must resolve that we will do one
of these two, and then stick to that resolution.

ISO 13250 presented one implicit model and a syntax for it. XTM 1.0
departed from both, and while I was unhappy with this I accepted it
because the new model and syntax were obviously superior to the first,
and because topic maps had not yet come into use to such a degree that
the change was much of a problem.

The situation now is entirely different, however, and anyone who
thinks that we can now change the model without high costs to the
vendors and users of the standard alike is deluding himself. We will
at the same time damage the credibility of the standard, and make it
suspect for a long time to come.

PMTM4 represents a revision to the XTM 1.0 model far more drastic than
the revision of the original ISO 13250 model in XTM 1.0. I believe
that it can be interpreted in a way that does not require the base
model to change, however. SRN supported this opinion when he wrote
that "PMTM4 is not about software"[1]; why his recent actions seem to
contradict that I am not sure.

Either we must make it clear that topic maps are still up in the air
and still under construction, so that nobody should use them for
mission-critical purposes, or we must make it clear that topic maps
are done, finalized for the foreseeable future, and ready for prime
time. To say that topic maps are now done is the same as saying that
the model is not going to change. The syntax is not so important, what
matters is the model.

Since the aircraft has just left the runway it seems to me dangerous
in the extreme to start unscrewing pieces of the structure, however
inappropriate their current positions may seem. At some point in the
future we are certain to want to adjust the structure, but to do so
before we have gathered experience with how the current structure
works seems to me inadvisable.

As I see it, this need not be so difficult. We have a model that is
accepted by the implementors (the infoset one), we have a model that
is accepted by the editors of 13250 (PMTM4), and we have a WG3
decision on how to proceed[2], which states that we should further
develop both, and produce a mapping between them.

I think we should agree to preserve the existing model, and move
forward according to the plan we already have.

--Lars M.

[1] I won't refer to the email in question. Ask SRN if you don't
    believe me.
[2] <URL: http://www.y12.doe.gov/sgml/sc34/document/0240.htm >