My talks at Extreme

[Murata Makoto <mura034@attglobal.net>]

I gave two talks in Extreme 2001.  In my first talk, I introduced classes of 
tree grammars (regular, single-type, local, restrained-competition), and explained 
which schema languages correspond to which class.  

My paper for this talks is available at:

[PDF] http://www.cobase.cs.ucla.edu/tech-docs/dongwon/mura0619.pdf
[RTF] http://www.cobase.cs.ucla.edu/tech-docs/dongwon/mura0619.rtf
[PS] http://www.cobase.cs.ucla.edu/tech-docs/dongwon/mura0619.ps

In my second talk, I introduced RELAX NG as a promising alternative to XML Schema 
and presented how DTD features are captured in RELAX NG.  Since people stick to 
inheritance, I mentioned my convention for mimicking inheritance.  I also mentioned 
relevant observations in my first talk (validation algorithms and boolean closure).  

Michael asked a few questions.  (1) Why does RELAX NG always use <ref> and <define> 
rather than introducing different mechanisms for referencing different things?  
[My answer: James think that one is simpler than many.]
(2) Isn't simple datatype assignment hampered by <choice><data .../><data ..../></choice>?
[My answer:  Yes, when you create such schemas.  Most schemas do not have this problem.]
Eve asked a few questions, but I'm afraid that I do not remember.  Somebody said to me that 
he was disappointed to hear that RELAX NG does not have inheritance, while others were 
very happy with conventions.

I believe that both of my talks worked well and that people in the traditional 
SGML community see advantages of RELAX NG.  However, since this community is 
merely a small portion of XMLers, we have to sell RELAX NG a lot more.

Cheers,

Makoto