On Thu, 2011-10-06 at 21:25 +1100, Michael Chapman wrote:
> On Wed, 5 Oct 2011, David Lutterkort wrote:
> > So far, nobody has needed that distinction, but it would be easy enough
> > to add a null function to the language so that you could pick out nodes
> > with null values, i.e. something like
> >
> > match test/*[null(.)]
> >
> > which, in your example, would match /test/b
>
> A null() function would be useful, however I'm not sure it should take a
> nodeset as an argument. It doesn't make sense for it to be applied to more
> than one node.
The reason I said nodeset is that there's no type internally for just a
node. There's two ways we can deal with passing more than one node: (1)
return true only if all the nodes have null values (2) throw an error.
I am inclined to do (1)
> Perhaps it should be a no-argument predicate, "this node has no value".
It would be more flexible if it takes an argument; I might do both, i.e.
make the argument to null() optional.
> Another question arises from all this: how would you match only non-null
> nodes?
Good point. We seem to be missing the boolean 'not' operator. I am very
tempted to do what XPath does and make not a function. With that, you'd
write
match /test/*[not(null())]
David
_______________________________________________
augeas-devel mailing list
augeas-devel@redhat.com
https://www.redhat.com/mailman/listinfo/augeas-devel
