Re: [xsl] Sequential numbers in pure xslt, breaking the no-side-effect rule

[Abel Braaksma <abel.online@xxxxxxxxx>]

David Carlisle wrote:
> > I would like to, but, I didn't invent the algorithm. The idea was this: 
> > make a UUID from within XLST 
> >     
>
> For Dave P? (Who asked me the same qn off list a while ago:-)
>   

Interesting. I did see the JUG attempts on the Saxon list, though.

> The functional way of doing it of course is to maintain a list of all
> the ones you've generated this session (or at least a count of the same)
> If the function took a sequence of strings (of uuids) and returned a
> sequence extended by one new one, then to generate three of the beasts
> you wouldn't do
> f(),f(),f()
> with side effects, you'd do
> f(f(f(())))
> which is so much more wholesome...
>   

Perhaps, but it requires two things: 1. input, 2. one location where you 
have to generate these IDs.

I don't have that luxury, I want it to be used 'out of the box', on 
separate places in the code and I want it to guarantee uniqueness. To 
quite some extend, I managed to do that, but I am missing the last bit.

But you do make an important point that I was overlooking. If I'd change 
the requirements to input an xs:integer that equals the amount of UUIDs 
needed, the user could create a pool of these UUIDs and use them.

It's not really what I wanted, but it is certainly the declarative way.

-- Abel