Re: Junit-type Framework for XSLT 2 Functions and Tem

On 5/18/05, David Carlisle <davidc@xxxxxxxxx> wrote:
>
> Dave, That  page doesn't claim that instancegenerator generates an
> infinite sequence of all instances, does it really do that?
>
> >    but I couldn't get it to work.
> >
> >    Instead, as a poor mans alternative.
> >
> >    1. From whatever source, use trang to go into rng.
> >    2. I have a stylesheet which then seeks all unique elements
> >     (and attributes), and can generate an xml file.
> >    It doesn't do the 'all instances' (which instance generator is
supposed
> >    to do), but it may provide a start Dimitre.
> >
> That sounds like a very finite list to me.
>
> To get the infinite sequence Dimitre asked for, of all schema valid
> documents ordered by length, note that any schema valid document is a
> finite string of unicode characters, and as the set of finite strings
> over a finite alphabet is countable, we can enumerate them in order.
> just list the natural numbers 1, 2, 3... in  base 1114112 (hex 10FFFF + 1),
> using unicode characters as the digits.
>
> That gives you the countably infinite sequence of finite unicode
> strings, then pass each string through a schema validator of your choice
> and discard it if it is not a well formed and schema valid instance of
> the schema in question. The resulting sequence is the requested result.
>
> I'm sure that Dimitre will be happy with the purity of the algorithm and
> not quibble unduely about its computational complexity or the expected
> time that it takes to generate any items in the sequence.


Sorry, I was absolutely stumped by your message and I understood what
you're saying only now.

I am accustomed to the lazy evaluation style of Haskell, where passing
parameters of infinite length is something natural,

like:

     take 5 [1,2 ..]

takes the first 5 elements of the infinite list of natural numbers.

It is most natural to take a finite number of elements (in our case
instances of a particular schema), from an infinite list generated by
another function.


Cheers,
Dimitre Novatchev.