Re: Type and Structure Re: ASN.1 and XML

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

On Mon, 4 Jun 2001, James Clark wrote:

> > Grammars should be an implementation technique,
> > not the nub of the question.

I just had a comment on this -- I will mention my view point -- I think
xml schemas have several properties that I think are useful for data
modeling -- I think even if in the past we did not have them, i think
these are *very* useful and convenience features -- one of them is closure
under union -- i think this is very useful.

also i think xml technologies are quite important and are slowly making
their inroads into academic books -- I think the undergraduate book for
formal language theory -- by Hopcroft, Motwani and Ullman, right now speak
of DTDs (they actually do not mention tree languages/grammars probably
because "unranked" tree languages were not the first to be studied, and
probably not as popular as we assume them to be, in stead they mention
them as context free languages and do *not* mention that DTDs are closed
under intersection -- anyways, these come later and have to be changed).

I think if xml schemas are to be understood in academia, they should be
taught as regular tree/hedge languages -- I think that can be the only
underlying principle.

regards - murali.