RE: Can XML Schemas Support Document Systems

[Murali Mani <mani@CS.UCLA.EDU>]

yes, one last mail from myself regarding XML Schemas' support to document
systems --

I think document processing community would like to move on w/o
considering other features -- for this, i think it is quite important to
know how to perform operations.

the remaining represent the difficulties we face if we use XML schemas --

If we do not have the 1-unambiguity restriction, then any model group and
regular expressions become *equivalent*. but if we do keep the
restriction, they are not -- in other words, the languages that are
1-unambiguous depend on the operators allowed by the model group.

So to understand what an operation can provide and how we can define an
operation, we should be able to analyze the model-group/regular expression
difference. There are 2 disheartening facts in studying this --
a) Every small modification in the model group definition is most likely
going to *severely* change our operation definition -- I think XML Schemas
has changed the model group definition so many times..
b) The regular languages that can be represented by 1-unambiguous model
groups has never been studied (even model groups as in SGML-- i think it
has only one additional operator -- any order or &). Actually Anne remarks
of the difficulty of this in 1998 (I this this is after almost a decade of
working in this area.) I do not think we have progressed any further...

So if non-determinism will not hurt anyone, i think it should be removed.
Also I think it is impossible to define and understand operations with XML
Schema. So I think it is *necessary* for us to have relax/trex
irrespective of whether we have xml schemas or not.

<warning>speaking for himself only</warning>

cheers and regards - murali.

On Fri, 27 Apr 2001, Bullard, Claude L (Len) wrote:

> The problem with this line of thinking is that XML
> has succeeded in removing means which
> one group was uncomfortable with and replacing it
> with means others are uncomfortable with.   I find
> examples where an author says "I want to
> work without DTDs, of course" then immediately
> begins to explain "generate-id" as begging the
> question of functionality.  As an SGMLer, it
> is strange to see that simple means
> one could quickly teach to a document manager or designer
> replaced with whole subsystems that
> require intense or shall we say, "extreme" skills.
>
> Spilt milk.  We need XML Schemas and need them now.
> The time for this debate was last year.
>
> So far, what I see in these arguments suggests that
> we can apply XML Schemas to document centric systems.
> Most of the counter examples seem to be extreme
> examples of the kind that don't often occur or for
> which there is a workaround.  No show stoppers.
> The gain of common means is worth the loss of
> fringe capabilities particularly if the fringe
> can be narrowed in future versions.  The question
> becomes one of controlling the process to ensure
> the working features don't change for the sake
> of competing notions and the fringe features
> are identified for true applicability so they
> do evolve sensibly.
>
> Len
> http://www.mp3.com/LenBullard
>
> Ekam sat.h, Vipraah bahudhaa vadanti.
> Daamyata. Datta. Dayadhvam.h
>
>
> -----Original Message-----
> From: Eric van der Vlist [mailto:vdv@dyomedea.com]
>
> The fact I have no SGML background and have taken for granted that I
> could work without DTD is probably a reason why I those constrains do
> not look "natural" for me.
>
> ------------------------------------------------------------------
> The xml-dev list is sponsored by XML.org, an initiative of OASIS
> <http://www.oasis-open.org>
>
> The list archives are at http://lists.xml.org/archives/xml-dev/
>
> To unsubscribe from this elist send a message with the single word
> "unsubscribe" in the body to: xml-dev-request@lists.xml.org
>