- From: Oliver Becker <obecker@informatik.hu-berlin.de>
- Date: Wed, 25 Feb 2004 08:58:23 +0100 (MET)
- To: Noe Michejda <noe_michejda@7thportal.pl>
- Cc: Ashok Malhotra <ashokma@microsoft.com>, public-qt-comments@w3.org
> Basically I have feeling, that both fn:substring-before and
> fn:substring-after
> should return same value when called with zero-length string
> - either zero-length string or unchanged first argument.
If you consider fn:substring-before('xxxxxxx','x') and
fn:substring-after('xxxxxxx','x') you wouldn't expect them to
return the same.
(Of course, otherwise we wouldn't need two different functions)
> Anyway I don't think, that logic of having 'virtual' empty string at
> beginning or end of every string should be determinant here.
I think it is important, and even more, it is logical and conistent.
Let fn:substring-after($a,$b) return $c
Let fn:contains($b,$bb) return true
Then I do expect that fn:contains(fn:substring-after($a,$bb), $c) is true.
Don't you?
In other words: if $c follows $b and $bb is in $b then $c follows $bb.
So we would have to argue whether fn:contains($a,'') should return true.
I think it should.
Best regards,
Oliver
(not a WG member)
Received on Wednesday, 25 February 2004 02:58:35 UTC