RE: Order By

I'm sure there are null values, and I'm sure that is
covered in the return results. What I seem to remember is 
that for null properties, when you ask for ALL the properties
of an object, you get the values of the non null properties
as XML elements, and you don't get any XML for null
properties. When you ask for a specific property of an object
that is null, you get an error that says it doesn't have
a value.

What I think you can probably get back for a zero length 
string is an XML tag with the name of the property and no 
data characters in it. I'm not 100% sure about that. I've been
away from the spec. for too long.

Whether or not that is true about null strings, 
just from a "principle of least surprise" point of view, 
one would expect that null strings would sort first,
because they are so similar to zero length strings, which
MUST sort first. So, it's a better design by the KISS principle. 
If null strings come first, that seems to settle the 
issue for all the other data types, again by the KISS principle.

Alan Babich

-----Original Message-----
From: Steve Haflich [mailto:smh@franz.com]
Sent: Thursday, June 08, 2000 7:12 PM
To: Babich, Alan
Cc: 'Kevin Wiggen'; www-webdav-dasl@w3.org
Subject: Re: Order By 


   From: "Babich, Alan" <ABabich@filenet.com>
   
   A null string is similar to a zero length string since
   neither contains any characters, so it seems odd to me 
   to put null strings and zero length strings at opposite ends 
   of the sort. I would think doing that could confuse some end users 
   who are not sophisticated in the subtleties of zero length 
   Strings Versus Null Strings. It also forces the implementation 
   to split the semantic hair of whether zero length strings are 
   null or not. ...

You still need to split that hair, methinks.  _If_ it is indeed
possible for an implementation to return both zero length and null
strings (is it possible???) then the sort order must specify which
sorts first.

Received on Thursday, 8 June 2000 22:33:47 UTC