[dsdl-discuss] Re: First draft of Part 7

Hi Eric,
 
Just a short reaction.

[EvdV:]
I am not familiar with Unicode enough to provide significant comments on
these points, but I don't see how we would attach these character sets
to the instance fragments. Could you elaborate on that?

[EvdV]
Do you have any idea where
http://dret.net/netdret/docs/wilde-crvx-www2003.html comes from and
shouldn't we take that into account as an input?

My first impression is that it can be complementary to what you've
defined: your proposal seems to be very focused on the definition of the
the character sets while this other one seems to be weaker in this area
but proposes methods to match the character sets to XML productions and
fragments.

[DG]
I did not work out the character restriction part yet, but what Erik Wilde describes comes closely to what I have in mind. Basically he says that the characters in all elements that match a certain XPath expression should be from a certain set, and what I add is that this set can have a name and that you can define your own private characters to be elements of that set; the named character set therefore acts more as a function, or better said, as a reusable definition. That is to say, a public named character set could specify all letters used in European languages by naming some code blocks and character classes. Then in your schema you could refer to that public definition, and by defining some of your private characters as having properties that would make it candidate for selection, those chartacters would then also be allowed. So at the moment you define the public named set, it still is a function. At the moment you "compile" the schema, you know what private properties you have added for your private characters, and then you can resolve the abstract named set definition to a set of Unicode chatracer points, including your own private characters.
 
I have corresponded with Erik Wilde on this subject, and he has some good ideas that I borrow, and some on which I have a different opinion. My very first draft had a reference to his work, but Martin pointed out that ISO standards only refer to standards.

I hope that next weekend (12-14 December) I can make some examples on how to restrict a part of a document, for instance an element and all its children, or all elements with an xml:lang attribute="nl" to a certain named character set, like latin-and-extensions or latin-dutch.

Best regards, have a good time in Philadelphia, sorry I can't join you,

Diederik

Ë™é›z»ŠÇ.²È¨žX¬N‹§²æìr¸›z™^jǬzwZ™ë,j°ŠØmr‰¦jwnžË›±ÊâmëhvÇev+ºË+z«ž²×lvZ+™¨¥¶‡lvWb±Ë¬²·ª¹ë-vÇe¢¸¹¸Þr
Received on Thu Dec 4 21:34:31 2003