RE: TREX and RELAX unite?

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

I have been studying schema efforts from July 2000. This actually is the
kind of conclusions I have come to at this point of time. I am very strong
and convinced of my conclusions. Also this happens to be a long mail.

Also because I was *very* doubtful of the standardization process, and
because I was interested not to use anyone's time on something when I
considered XML Schema might be adopted as the standard by W3C, I tried to
ensure that I do not involve too many people (even professors who were
interested) in this study -- even for projects, I debated long about what
is the technology which fellow project members should know. But from the
little talks I could manage with the professors, they were *really*
surprised when they saw that the competitors are RELAX and XML Schema --
the choice was *very* obvious to them.

I am biased towards RELAX and TREX and am biased *against* XML Schema. I
think my bias is based on which approach is more correct -- this could be
subjective, but I feel RELAX and TREX are correct.

The statement James Clark is an expert I think is an understatement by all
means -- I am sure there are many people more aware of XML research than
me -- but from what I have seen, James Clark has the cleanest
specifications for XML -- XPath 1.0, XSLT, the best implementations --
SGML parser, XT for XSLT, the best algorithms for regular tree grammars
(or TREX or RELAX).

Also I believe Makoto Murata was the *first* person to mention that
closure properties are important and requested the schema WG to consider
them, but I think this has been ignored by the schema WG for whatever
reasons.

Also, I am very greatly thankful to James Clark and Makoto Murata for the
efforts they have put in to materialize this effort on schema. I think
this effort ensures that XML research is going in the right directions.
Also more the people help them with the development of approaches like
RELAX or TREX, I think faster this approach will mature.

I do not say I am fully confident -- but again, this is something which I
said before -- I think RELAX and TREX are the way schema languages should
be if we want to do document processing such as XSLT or XQuery. Again we
learn a *lot* from XML Schema such as inheritance and constraint
specification such as key constraints.

regarding the difference between RELAX and TREX, whether it is tree
regular expressions or hedge automata, I think we as users need not be
aware of that -- it is something I believe will *not* reflect on users who
wish to use this approach for schema -- I think regular tree grammars,
regular tree expressions, hedge regular grammars, hedge automata
essentially are all equivalent, the only difference is in there are
multiple algorithms for doing validation based on these very well studied
(in the past) approaches.

And as the articles says, i also agree that schema is a quite complex
problem, where not all issues were visible at the beginning, but I think
the core and important math assumes a slightly less important role in the
present W3C spec.

<warning>speaking for himself only</warning>

regards - murali.

On Mon, 23 Apr 2001, Bullard, Claude L (Len) wrote:

> This comment needs clarification:
>
> "The main problem I have with it is it's too complicated," said James Clark,
> an XML expert in Bangkok, Thailand, who launched his own schema effort,
> called Trex, in December. "It's fine if you're a huge company like Microsoft
> [Corp.] or IBM. They can just add more developers."
>
> Is Clark comparing the developers of schema support tools (eg, schema
> engines) or authors of schemas?  Presuming the former, from the
> perspective of tool users, is this as interesting an issue as the
> possibility of having several schema-type alternatives?
> Intertransformability will be the issue.  It's ugly but it is
>
> a way to enable choice among technical means.
>
> Len
> http://www.mp3.com/LenBullard
>
> Ekam sat.h, Vipraah bahudhaa vadanti.
> Daamyata. Datta. Dayadhvam.h
>
> -----Original Message-----
> From: Michael Champion [mailto:mike.champion@softwareag-usa.com]
>
> eWeek has an article about the W3C XML Schema complexity controversy at
> http://www.zdnet.com/eweek/stories/general/0,11011,2710691,00.html
>
> ------------------------------------------------------------------
> 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
>