RE: [xml-dev] a namespace definitions related question(s)

["Shlomo Yona" <S.Yona@F5.com>]

Hello,

Thank you.
Conformance to the standards and recommendations is indeed important to me.

Hopefully, as my understanding of the standards increases so will the relationships among them will get clearer.

Thanks again to you and others that have contributed to this discussion. It was useful and informative to me.

Shlomo.

-----Original Message-----
From: noah_mendelsohn@us.ibm.com [mailto:noah_mendelsohn@us.ibm.com] 
Sent: â 20 ôáøåàø 2007 19:07
To: Shlomo Yona
Cc: Pete Cordell; xml-dev@lists.xml.org
Subject: RE: [xml-dev] a namespace definitions related question(s)

Shlomo Yona write:

> What I'm asking is, is there a downside to "semantically" 
> matching start/end tags by using namespace information?

Yes indeed, there is a big downside.   Part of the value that XML brings 
to the industry is good, uniform interoperability.  What's accepted by one 
XML processor is generally accepted by another.  That's true across 
software vendors and hardware platforms.  The same XML documents will 
likely be successfully readable by pretty much any commercial or open 
source XML database, by a range of "office" applications, by XML editors, 
by XSL-based styling tools, etc., etc..   While there can be sort of 
oddball reasons for occasionally using software that accepts input that 
does not conform to the specifications, proliferation of software that 
fails to enforce the specifications encourages people to write and 
exchange buggy XML documents.  That in turn puts pressure on those who 
have written "correct" software to modify it to accept whatever sorts of 
documents seem to be flying around.  But, and this is the really big 
problem, there's no specification for such looser content, so there are no 
test cases, so everyone's likely to do it differently.  The 
interoperability that made XML valuable in the first place is lost.

Now, if you're asking:  if the XML and Namespaces Recommendations had been 
developed together, rather than one after the other, might there have been 
some other options open for tag matching?  Yes probably.  But that's not 
how the specifications were developed.  By the time Namespaces were being 
crafted, XML was already a stable Recommendation.  The designers of 
Namespaces made the call that their specifications would apply only to 
documents that were already legal XML, and that meant that they could not 
even seriously consider matching tags at the semantic level you're asking 
about.

Given that the specifications are as they are, I'd strongly urge you to 
write content that conforms, and to use software that enforces such 
conformance, wherever reasonably practical.  The payoff for you is that 
your XML will work with an extraordinarily broad range of current and 
future software.  They payoff for everyone else is that none of our 
support lines will be ringing with bug reports like:  "my XML works in 
Shlomo's software, how come it doesn't work with yours?".

--------------------------------------
Noah Mendelsohn 
IBM Corporation
One Rogers Street
Cambridge, MA 02142
1-617-693-4036
--------------------------------------