[topicmapmail] subject proxies and subject maps

[Lars Marius Garshol]

* Johannes Schmidt
>
> The interpretation of that lonely proxy $sp as a subject map with one
> proxy is not "in" the application. It is just my interpretation (or
> someone else's). I am not sure if the "problem" is clear. $sp is
> independent from any map and legend, it only exists in the application
> context. So - for example - no navigation operation can be applied as
> navigation operations are relative to subject maps (which means that
> this proxy is inaccessible for a TMQL retrieval, there is only
> application specific access via labels or ids or similar).

This is all true. However, you can perfectly well implement this in  
such a way that a proxy *is* bound to a specific map. Or, you can  
choose to define your navigation operations in such a way that they  
require two parameters: the proxy and the map. If you stop to think  
about it, I think you'll see that this is actually what TMRM does.

> More abstracted, the question could be: Are there elements without  
> sets?

Yes. (Does the number 1 exist independently of the set of all  
integers? Do you exist independently of the set of all German people  
who have heard of Topic Maps and can speak English?)

> It seems to be useful to implement the interpretation of a lonely  
> proxy
> as a subject map with one proxy explicitly. If not created "in" a
> subject map (whatever this means) put it in a primitive map (which  
> then
> has to be created) without a legend (or provide that "noop" legend)  
> and
> basic ontology. Labelling has to be done in any case.
> This would mean: There is no element without a set.

You can do this. There's nothing in the TMRM that forbids you from  
doing this. In fact, you'd be hard put to find *anything* which the  
TMRM forbids you from doing.

The questions should probably be: what am I trying to provide to the  
user of this API? How should the API work to best support that user?

--Lars M.