Frederic,
On 2005 Jun 2 , at 18.24, Frederic V. "Rick" Hessman wrote:
> The point is that the concatenation of UCDs in its present form (using
> a
> semicolon) is effectively an array of UCD strings. The question is
> only
> which character is defined as the separator. Since blanks are not
> allowed
> in UCDs, then a blank is as good a separator as a semicolon. Our
> argument is that - from an XML perspective - blanks are MUCH BETTER
> separators, since blanks permit us to check the syntax of a UCD entry
> in principle whereas a semicolon-separated array of UCD strings looks
> like a simple string to XML parsers.
UCDs are orthogonal to XML, thus changing the UCD syntax because one particular XML technology allows you to peek inside attribute values (which are opaque in the XML data model), is the tail wagging the dog.
Also, any meaningful checking of UCDs is going to have to do more than split the UCD at semicolons, so having your XML parser API do that for you doesn't seem much of a win.
Blanks would also be a bad intra-UCD separator, since that would stop blanks being the natural inter-UCD separator. With the semicolon, one can naturally talk of a space-separated list of UCDs, with whatever meaning is appropriate in the context; without it, one cannot.
All the best,
Norman
-- ---------------------------------------------------------------------- Norman Gray : Physics & Astronomy, Glasgow University, UK http://www.astro.gla.ac.uk/users/norman/ : www.starlink.ac.ukReceived on 2005-06-03Z10:10:23