W3C home > Mailing lists > Public > www-math@w3.org > July 2010

Re: How to use CDBase URIs in Content MathML?

From: David Carlisle <davidc@nag.co.uk>
Date: Mon, 12 Jul 2010 18:10:20 +0100
Message-ID: <4C3B4C7C.9070803@nag.co.uk>
To: Urs Holzer <urs@andonyar.com>
CC: Christoph LANGE <ch.lange@jacobs-university.de>, "www-math@w3.org" <www-math@w3.org>
On 12/07/2010 16:41, Urs Holzer wrote:
> I wrote as answer to David:
>> Isn't this the same as claiming that there is no need for a namespace
>> mechanism for XML because elements would have unique names anyway?
>
> Please don't consider this answer as offensive.

Ive been in too many email discussions to be offended by them:-)

 > Perhaps I misformulated
> it a little. The substring "OM" in all the element names from the
> OpenMath XML encoding are there because at some point, it was not
> possible to just use an XML namespace.

At some points in the history of Openmath XML didn't exist, nevermind 
xml namespaces:-)

> Let me show the parallels of the
> cdbase and the XML namespaces:

I believe that the OM archives from around 1999 will show some 
presentations of mine giving that analogy yes.
>
> In my opinion, the prefix of an element or attribute in XML is similar to
> the OpenMath cd. The local name in XML is similar to the symbol name in
> OpenMath. Finally, the URI of the namespace is similar to the cdbase in
> OpenMath.

OpenMath explictly decided _not_ to formalize such an analogy (which 
would have made the cdname a local arbitrary name in the manner of an 
xml namespace prefix) so whil ethere are some similarities, it's best 
not to push the analogy too far.

> It is widely accepted that just chosing prefixes carefully in XML without
> using URIs to identify the namespaces is not enough in practice.

widely but not universally, note for example the current html5 proposals 
more or less completely reject the namespace/uri extension mechanisms in 
favour of centralised name registries. Not that I necessarily agree with 
that, just observing that (especially in a web context) using URIs as 
naming systems does not always help adoption.

 >  I
> believe that the same is true in the case of OpenMath and therefore also
> Content MathML: Chosing the cd carefully is not enough in practice.
>
> Greetings
> Urs

naming registraries can work, eg ctan acts as one for tex and in 
practice if you go \usepackage{longtable} you always get the same 
package. But as I said in my initial reply to Christoph, if you do want 
to consider the cdname as a local name that might clash, this is still 
OK as long as you take a sufficiently relaxed interpretation of the 
cdgroup file to allow you to use the CDURL elements there to give you a 
URI based name disambiguation.

David
Received on Monday, 12 July 2010 17:10:51 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Monday, 12 July 2010 17:10:51 GMT