Sam, Since you're not 1.0, and theoretically people shouldn't be relying on this behaviour in your implementation, the answer is, there's no compelling reason to support it, but it's a nice to have in case anyone actually was relying on it. I just piped up because I thought it was an "obvious" alternative, and works out better in that the localPart wouldn't end up in an illegal state through attempting to use the old Axis form.
--Wes -----Original Message----- From: Sam Ruby [mailto:[EMAIL PROTECTED]] Sent: Thursday, June 20, 2002 12:10 PM To: [EMAIL PROTECTED] Subject: Re: [jsr110-eg-disc] QName (COMMITTERS PLEASE READ) Wes Moulder wrote: > Ya'll might not like me stepping in here, but I have two questions > about this: > Hi Wes! You are always welcome here. > a) isn't a local part of "one:two" illegal? > I would think so... > b) Why not make your QName do both from the valueOf() method? > I'm a bit concerned about how code that depended on this feature would react if, some how, another implementation of this interface were to be picked up in the classpath. Is there a compelling need to support this other syntax? > It should be trivially easy to figure out which one of the two forms > the string is in, and since a : is illegal in an NCName (what the > localpart is defined as), it shouldn't be difficult to figure out > which form it's in. > I agree that technically it would not be difficult. - Sam Ruby
