[sc34wg3] Semicolon or not semicolon

[Lars Heuer]

Hi Dmitry,

[...]
>>And if that's something for CTM and we want support it directly (not
>>via the work-around), we could create a rule like this:
>>
>>    ctx-binary-assoc ::= <topic-ref> <topic-ref>
>>
>>Within the "topic" production without changing CTM too much. Maybe it
[...]

While thinking harder about it, I recognised that this syntax wouldn't
work without problems since we're just naming all identities

     john works-for The-Beatles .      # Works
     john o:works-for The-Beatles .    # May work, but requires some
                                       # work
     john o:works-for o:The-Beatles .  # Does not work

The problem is that o:works and o:The-Beatles may be recognised as
subject identifiers which belong to the topic "john".

Hmm... I think, I understand the never explained "binding identities
to local identifiers" now. Iff we use a prefix we can decide if
something is meant as assoc type or as additional identity. I wish
Steve or you wouldn't have kept this secret.

    john ~ o:works-for  # subject identifier
         ' o:something  # item identifier
         = o:bla        # subject locator
         o:works-for o:TheBeatles    # Binary assoc.
         plays(o:piano, o:Imagine)   # Tpl. invocation
    .

If a topic has more than one identity, all additional identities
require a prefix. Taaadaaaaa! :)

The remaining question is, if the subject/object stuff is a useful
addition to CTM.

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