Re: [xml-dev] Most XML vocabularies are too large and inevitably havelot

[James Fuller <james.fuller.2007@gmail.com>]

On Sat, Dec 17, 2011 at 8:50 PM, Costello, Roger L. <costello@mitre.org> wrote:
> Hi Folks,
>
> Recently I have been learning Lambda Calculus [1].

good man

> A fascinating thing about Lambda Calculus is its richness, despite it being extraordinarily simple.

and most of what follows I did not follow ...

I would start off with what I think seems to be your fundamental
assumption e.g. that Lambda Calculus is correct and formal as a direct
result of its boundedness and 'smallness'.

Whilst Lambda Calculus maybe true and obvious I would not be so bold
as to then vary this with something else. I am not going to argue
against your fundamental assumption but lets remind ourselves that
math has a lot more holes in it ... I find the Sorites paradox[1] a
good reminder of just how fragile simple assumptions can be blown away
in math, tout de suite.

Data and code (and things like XML Schema) are like languages because
humans create them, but they are extremely abstract and do a very poor
job at nailing the jelly to the ceiling; I don't want to be an XML
Schema apologist ... I think the latest version makes it much more
usable then before and I am grateful for the work that went into it; I
think we all agree something better is possible but I will be damned
to know if that will occur in my lifetime (if there is an example in
any programming domain of something good, then send it my way).

I agree with you that xml vocabularies that are small are probably
more successful, just do not agree with you your reasoning. I don't
agree that the number of terminals, operators or 'things' in the
language can always result in incomprehension or complexity ... think
CISC vs RISC, depending on your current want/need you may find a more
complex instruction set better.

I have far too little experience to authoritatively comment on
research topics in schema but just as coding in the language of the
domain is a good thing and encoding your data in the language of the
domain is a good thing ... it maybe that schema approaches that are
generic will always create unwanted complexity ... would be great to
hear from gurus their opinion of a successful schema tech over the
past 20-30 yrs.

as always interesting topic

Jim Fuller

[1] Sorites paradox - http://en.wikipedia.org/wiki/Sorites_paradox