Re: Atomization in Unary Lookup

On 11/05/2016 10:44, Michael Kay wrote:
> Well, (a) that's clearly the way the spec is written, and (b) I doubt anyone considered the alternative before, and (c) the alternative is probably acceptable, but is it worth changing?
>
> It's worth adding a test case though.

I notice that in SaxonHE, the result of

[1, 2, 3]?[1,2]

is

(1,2)

i.e. it has done

for $k in fn:data([1, 2])
return .($k)

I suspect that at this stage, it really would be wiser to make the 
specification match this behaviour, otherwise the type checking can get 
quite hairy.

I will raise a bug report.

Cheers,
     Tim

Received on Wednesday, 11 May 2016 10:18:45 UTC