Re: [xsl] Are these equivalent XPath expressions for determining that there are no duplicates in a list of items?

[Michael Kay <mike@xxxxxxxxxxxx>]

As David says: there's the "//" difference, and there's the fact that 
index-of() has slightly different comparison semantics - "=" throws an 
error if you compare a number to a string, index-of() effectively 
returns false.

Both are O(n^2), and are therefore poor solutions given that there's a 
perfectly good O(n log n) solution, namely

count(Websites/*) = count(distinct-values(Websites/*))

Michael Kay
Saxonica

On 01/11/2012 13:01, Costello, Roger L. wrote:
> Hi Folks,
>
> Are there no duplicate websites in this list:
>
>      <Websites>
>          <Website id="Amazon"> http://www.amazon.com </Website>
>          <Website id="Apple"> http://www.apple.com </Website>
>          <Website id="Ebay"> http://www.ebay.com </Website>
>          <Website id="Google"> http://www.google.com </Website>
>          <Website id="Microsoft"> http://www.microsoft.com </Website>
>          <Website id="VirginAtlantic"> http://www.virgin-atlantic.com </Website>
>      </Websites>
>
> Here is an XPath 2.0 expression to answer that question (thanks to Dimitre):
>
>      empty(Websites/*[index-of(../*,.)[2]])
>
> Here is an XPath 1.0 expression to answer that question:
>
>      not(//Websites/*[. = preceding-sibling::*])
>
> Are there any situations where those two XPath expressions differ in their answer?
>
> /Roger