[sc34wg3] Fwd: Re: CTM: prefixes and include

[Lars Heuer]

Hi Robert,

[...]
>> Your note is correct. The "included" file is invisible for the file
>> which imports it. If file A includes file B, only the templates from
>> file B are are visible for file B, the rest (%prefixes etc.) is
>> invisible.

> Is this somewhere addressed in the current draft?

Thought, that this is implicit. The document which includes the other
document, does not "see" the prefixes since the other document is
deserialized and converted into Topic Maps constructs.

After reading the section again, I found a failure: It is not clear
to which prefixes the QNames within the templates refer to. They
should refer to the prefixes declared in their own file.


[Named wildcards]
>> Good question, this might be underspecified in the current version. :(
>> ?xxx in file A should merge with ?xxx in file B, but I am afraid that
>> they do not merge currently. yyy in file A merges with yyy in file B,
>> that should work, but I am not sure if it works with named wildcards.

> I somehow had the intuitive feeling that ?xxx in A and B should be NOT
> merged. The named wildcard thing should be kept as local as possible,
> IMHO.

IMO they should be merged since the include directive is meant for
situations where the author controls both files.

And the merging-policy supports the proposal to handle the include
directive like a text replacement (the included file is inserted into
the current stream). But inserting the included file into the stream
does not work, therefor we need the item identifiers on topics to let
the topics in file A and B merge; so the wildcards should be handled
the same way.

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