[sc34wg3] CTM - Open Issues

[Lars Heuer]

Hi all,

Here a list of open issues for CTM.

Any input (proposals, critics etc.) is highly appreciated. :)


Item Identifier marker
----------------------
We need a marker for item identifiers since "~" is used by TMQL (as
suffix) for subject identifiers and by CTM as reifier marker.

Proposals:
a) Keep "~" as item identifier marker and change the CTM reifier marker
   to "<~".
b) Invent another item identifier marker, i.e. "`" or "'" and keep the
   CTM reifier marker.
   "`" may cause problems because it is too subtle.
c) Use "#" as marker for item identifiers and invent another symbol to
   introduce single line comments ("#" indicates a comment in the
   current TMQL / CTM drafts)


CTM-wildcard vs. TMQL tm:subject
--------------------------------
TMQL uses the "*" as reference to the topic "tm:subject", in CTM we
use "*" as instruction to create a topic and *foo as named wildcard.

Proposals:
a) CTM uses "?" as instruction to create a topic. And ?foo to create a
   topic to which can be referenced to (named-wildcard).
b) CTM uses "_" as instruction to create a topic and "_foo" as named
   wildcard (this proposal violates currently our definition of
   identifiers, since "_foo" is a valid identifier. We have to limit
   the possible identifiers to implement this proposal)
c) CTM uses "_:" as instruction to create a topic and "_:foo" as named
   wildcard
d) This is no issue, the difference should stay


Multiline Comments
------------------
TMQL and CTM do not provide multiline comments. Some think, both
should support these.

Proposals so far:
a) #(  comment )#
b) (-: comment :-)
c) (o- comment -o)
d) (: comment :)   (c.f. XQuery)
e) #/ comment /#


Proposal to allow embedding IRIs in <>
--------------------------------------
Currently, TMQL and CTM provide only plain IRIs and the user must take
care if the following character may be interpreted as part of an IRI
or not (c.f. <http://www.jtc1sc34.org/repository/0905.htm#sec-assoc>).

Proposal:
- CTM and TMQL should allow to embed IRIs within angle brackets (<>)
  Reasons:
  - RFC 3986 recommends the <> syntax for IRIS
  - template-name(http://example.org/ , "arg") looks odd and humans
    are conditioned to leave a space *behind* a comma but not before a
    comma
  - Providing CTM examples via e-mail may result in invalid CTM
    fragments since e-mail clients may break up long IRIs (they do not
    break IRIs embedded in <>)
  - Common syntax for IRIs
  - Does no harm, if we allow both syntaxes (either embedded or a
    plain IRI)

Regards,
Lars
-- 
http://www.semagia.com