[sc34wg3] TNC backward compatibility

[Lars Marius Garshol]

* Nikita Ogievetsky
| 
| Hmmm... I see your point. But standard said that TNC is always ON.
| As far as I understand, that dictates what is backward compatible
| with the standard and what is not.

Perfectly true, but we know for a fact that this interpretation will
break existing data. In that case I feel it's better to adjust to
reality and choose the least troublesome default. Also, unless we do
this we haven't actually removed this bug at all, so there's a huge
difference between ON and OFF here. Setting it to ON effectively means
we're still stuck in the mud.
 
| I mean that one can now use occurrences instead of subjectIdentity.

Yes. However, the semantics are different, as they always were. So
effectively the temptation to abuse subject indicators has now become
far less, which is a good thing.

| The distinction becomes very fragile and scary.

Not at all. It's a clear semantic difference, like that between base
names and occurrences.

| For example one can use occurrence of type "subjecIndicator".
| BTW, in this case we are getting very close to my RDF Topic Maps schema.
| But I am not sure that we want to go there. 
| It may change the way people work with Topic Maps.
| May be I am exaggerating?

I think you are. :)
 
| That is why some time ago I said that we are about to open a Pandora
| box if we do not think about this very carefully now.

By all means.

-- 
Lars Marius Garshol, Ontopian         <URL: http://www.ontopia.net >
GSM: +47 98 21 55 50                  <URL: http://www.garshol.priv.no >