[XML-DEV Mailing List Archive Home] [By Thread] [By Date] [Recent Entries] [Reply To This Message]

Fwd: infoset data model

• To: xml-dev@l..., Erik Wilde <net.dret@d...>
• Subject: Fwd: infoset data model
• From: "Simon St.Laurent" <simonstl@s...>
• Date: Sat, 17 Aug 2002 08:55:16 -0400

Yes, Erik - you do need to be subscribed to this list to post messages.
(There are still some problems for subscribers, but that's another
story.)  

I'm forwarding this to xml-dev; I guess people should copy Erik for now.

====== Forwarded Message ======
Date: 8/17/02 11:46 AM
Received: 8/17/02 5:48 AM
From: net.dret@d... (Erik Wilde)
To: jcowan@r... (John Cowan)

hello.

is there any particular reasons that my posts do not appear on xml-dev? 
is it necessary to be subscribed to this list? i'd rather not, because 
there is just too much volume and noise on this list.

to get to the point:

it is very unfortunate that the discussion seems to drift away in a 
number of directions now, including xlink 2.0 requirements. do you have 
any idea how to keep the discussion focused? otherwise, i think it's 
interesting but i will not have any consequences...

to really get to the point:

John Cowan wrote:
> Erik Wilde scripsit:
>>i think before we can define a generic infoset serialization, we need
to 
>>have a proper data model of the infoset. the infoset spec is extremely 
>>vague about what can be done, for example it does not even say that 
>>there are only items and properties. and it does not say which values 
>>properties may have, and whether there are some universally understood 
>>special values (such as "unknown" and "no value" as they appear in the 
>>infoset spec).
> Well, there are only items and properties.
> Every property (I believe; if not, please raise an erratum) has a type
> given, either "set of items", "list of items", or a simple type.
> "No value" is a special value.  "Unknown" is used when we are dealing
with
> partial infosets.

is this your understanding or really explicitly stated in the spec 
(forgive my ignorance, it has been some time until i really read the 
spec). anyway, i think we *need* to make some clarifications:

- items and properties are the only first-class constructs

- properties may be sets, lists, or simple values (does that remind you 
of xml schema simple types with their list and union types... ;-)

- what are the simple types? are they xml schema simple types? if not, 
what else? if so, how do i specify these types? the infoset spec 
currently does not contain any formal type definitions.

- how can we combine infosets models in a modular way (for example, base 
xhtml on the xlink infoset which is based on the xml core infoset)? can 
we identify infoset models with a namespace uri?

just yesterday a number of new xquery drafts have been released. i think 
we should try to introduce the concept of infoset extensions there, in 
particular in xslt/xpath (i don't know enough about xquery), so that 
infoset extensions can be accessed via xpath extension axes (identified 
by namespace prefixes). if properties indeed are based on xml schema 
simple types, then we will not get into trouble with xpath's data model 
with this (otherwise, we will get into really deep trouble).

cheers,

erik wilde  -  tel:+41-1-6325132  -  fax:+41-1-6321035
           mailto:net.dret@d... -  http://dret.net/
           computer engineering and networks laboratory
           swiss federal institute of technology  (eth)
           * try not. do, or do not. there is no try. *

====== End Forwarded Message ======

-------------
Simon St.Laurent - SSL is my TLA
http://simonstl.com may be my URI
http://monasticxml.com may be my ascetic URI
urn:oid:1.3.6.1.4.1.6320 is another possibility altogether

• Prev by Date: RE: Re: Can XLink be fixed?
• Next by Date: RE: Re: Can XLink be fixed?
• Previous by thread: Do sheep dream of electric URLs?
• Next by thread: linking, 80/20
• Index(es):
  • Date
  • Thread