[sc34wg3] Embedded topics as association type?

[Lars Heuer]

Hi all,

[...]
>> Seriously: As we introduced embedded topics we agreed that we do not
>> want them at the top level in a CTM document. But if we allow embedded
>> topics as association types, we'd have embedded topics at the top
>> level of a CTM file.

> Ahhhhh. Right. Yes. Indeed.

>> We both talked about it and we agreed that we shouldn't allow  
>> embedded topics as association types because users might create  
>> association types without an explicit identity:
>>
>>      [- "member of"](member: john, group: The-Beatles)

> True, although not sufficient reason in itself. But the above was  
> indeed the reason.

Okay, a funny discussion about our supermodel status and the embedded
topics led to the question if shouldn't allow embedded topics for
association types even if this is bad practise if the user does not
specify an explicit identity. But CTM already allows association types
without an explicit identity using wildcards:

       ?(member: john, group: The-Beatles)

       ?foo(member: paul, group: The-Beatles)

If nobody objects I'll allow embedded topics as association type in
the next draft just to get a more consistent CTM grammar and a uniform
policy for embedded topics.

Best regards,
Lars
-- 
Semagia 
<http://www.semagia.com>