[dsdl-discuss] Re: Comments on Hull and Kernel RNC proposals

Martin,

> 1) Collections need to be assigned a user readable name and a unique
> identifier that can be used in the URI of the href attribute of the ref
> element to ensure correct identification of groups.

Is <ref href="http://someReadableDomainName/someReadableDirName/someReadableCollectionName"/>
good enough?

> 2) There needs to be some way of adding a named private use character to
> a collection. The adopted technique should allow the private use
> character to be defined as a member of an existing class, such as
> lowercase character, in such a way that it will be added wherever that
> class is referred to.

It is already possible to name a collection that consists of private
use characters. It is also possible to create a collection as the union
of existing collections and prive use character sets.

Some people might expect that DSDL should allow users to change
the Unicode database and thus control formatters or rendering engines.
But I very strongly oppose to that approach, since I believe that
they are not in the scope of SC34. Even if we try, I do not believe
that Unicode implementors will care what we have to say.

> 3) The definition of collection, hull and kernel as a single coll seems
> to allow kernels within hulls, hulls within kernels and even hulls
> within kernels within hulls within kernels..... What is the union of a
> kernel and a hull? What is the difference between two hulls, two
> kernels, etc? The ways in which the current definitions nest infinitely
> needs review. There needs to be some way of defining a collection that
> is based on sets of multiple, non-nested, kernels and hulls that combine
> requirements for different languages.

This is already explained in the original proposal by Martin Duerst. In
the CD, I will borrow his wording and make sure that the semantics is clear.

> 4) If, as you state "The semantics of [29] thru [37] depend on the
> version of the Unicode standard" then there must be a way of specifiying
> which version of the Unicode standard is being referred to. There should
> also be some way of referring to the names used within a particular
> version of a standard to identify named sets of Unicode characters.

I agree that there should be some mechanisms for specifying
Unicode versions. IsBlock (defined by [36]) allows the use
of named sets of Unicode characters defined by the Unicode standard.

> 5) It is not clear why the only place you can define a charClassExpr is
> as a qualifier to a charClassSub definition. Why can't charGroup
> elements be defined using charClassExpr directly? Why are there two
> definitions called charGroup?

I just copied these production rules from XML Schema Part 2. If we
change, we cannot use existing implementations of XML Schema Part 2
as a basis for Part 7 implementations.

> We look forward receiving a more detailed draft in which the semantics
> of the proposed constructs are formally defined in the near future. A
> recommendation to this effect will be issued at the closing plenary.

I will create a CD in the near future. (But I will first finish Part 4.)

Cheers,

-- 
MURATA Makoto <murata@hokkaido.email.ne.jp>
--
DSDL members discussion list
To unsubscribe, please send a message with the
command  "unsubscribe" to dsdl-discuss-request@dsdl.org
(mailto:dsdl-discuss-request@dsdl.org?Subject=unsubscribe)