W3C home > Mailing lists > Public > www-math@w3.org > September 2007

Re: Wish for MathML 3: version attribute

From: Ray Kiddy <ray@ganymede.org>
Date: Mon, 17 Sep 2007 08:40:22 -0700
Message-Id: <18E99F93-CA93-4E05-91FA-04E455804F2C@ganymede.org>
To: www-math@w3.org

On Sep 17, 2007, at 7:08 AM, Michael Kohlhase wrote:

>
> I like this, let's continue.
>
> Michael
>
> David Carlisle wrote:
>>>   dtd="important" >
>>>
>>
>> It would be nice if it were possible to flag a fragment as  
>> depending on
>> a dtd or not but i fear this isn't possible.
>> If the fragment uses entity references (so requires a DTD) then
>> (assuming that it is read by an XML parser rather than custon  
>> software)
>> it will have been rejected with a fatal error if there is a  
>> problem with
>> the dtd during the pase stage so (conceptually at least) before  this
>> attribute value  can be inspected by a mathml application.
>>
>>
>>> <math version="MathML3.0", "p-MathML3.0+OpenMath2.0"  ...
>>>
>>
>> Hmm interesting. Allowing this would mean (at the specification  
>> level)
>> that the version text would be arbitrary string, with certain  
>> communities
>> coming up with conventional values for profiles that have common  
>> usage?
>> Clearly there are uses for that although the advantage of having a
>> strictly numeric value is that you can do numeric tests on it (cf the
>> forward compatible processing rules for xsl:version).
>>
>> Perhaps separate out teh numeric version and a string based profile:
>> version="3.0" profile="presentation mathml with OOMML attributions"
>> being a profile of topical interest....?
>>
>> David

I am not presuming to comment on the rest of the discussion, but from  
a data categorization viewpoint, this last comment poked some things  
in my brain.

What people end up wanting in this kind of "profile" is flexible  
categorization. This can be provided if the profile is a string  
specified to be of the form:

	profile="<string>"  |

	profile="<key><sep1><value>"   |

	profile="<key1><sep1><value1><sep2><key2><sep1><value2>..."

The two <sep> characters should not appear in <string>, <key>, or  
<value>. <sep1> can equal <sep2>, if that is specified. If these  
formats are specified, a lot of thrashing about further down the road  
might be avoided. Finding appropriate separator characters after the  
fact can be challenging.

Just a suggestion.

thanx - ray
Received on Monday, 17 September 2007 16:13:24 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Saturday, 20 February 2010 06:13:00 GMT