Re: Last Call Working Draft of MathML 2.0, 2nd edition published

Bill,

Thanks to your observations, we have added the definiton specific 
information to the URI.  The change should mean that the link goes 
straight to the exact definition instead of  leaving the reader to 
deduce which of the several definitions found at that the former URI 
they should use. 

However, in general, there is no requirement  that this be a link to a 
webpage, and it is beyond the scope of MathML2 to provide guidelines on 
how to structure the URI or to try to desribe an OpenMath dictionary.

Here are some things to consider.

   1. There is nothing to prevent the author from providing just a
      name.  Such a name still warns the reader (including any software
      consuming such an expression) that a symbol is being used in some
      special way.

   2. The definitionURL must also allow an author to identify "any" 
      reference including, for example, a definition found only in some
      classical manuscript located in the physical archives of some
      library. This could even take the form of mentioning the function
      name and giving the traditional bibliographic reference - all
      properly encoded as a URI, of course  

   3. The mechanism is supposed to work even if you have to contact
      OpenMath (or the library) directly to find out what it means. 
      Technically this could even take the form of 
      "author-definition2003-12"  ( and perhaps a phone number :-)   
      ).  Of course, good choices will make for better archival value.

It is the responsibility of the author to provide enough information to 
locate the required definition either through some obvious  embedding of 
the information in the string or through an actual link.  

The architects and authors of repositories such as OpenMath can (do) 
help considerably by providing authors
with standard mechanisms to make such references but must be free to 
design an construct those indexes according to their requirements.

Largely  because of these points we decided that it was best to leave 
the description of  any  such indexing mechanisms to the respositories 
themselves and to just include examples.

Once again, to help us put together our last call report, we would 
appreciate it if you could post a brief message acknowledging we've 
responded to your comments.

Stan Devitt.

Bill Naylor wrote:

>On Sun, 8 Jun 2003, Stan Devitt wrote:
>
>  
>
>>Date: Sun, 08 Jun 2003 13:14:54 -0400
>>From: Stan Devitt <jsdevitt@stratumtek.com>
>>To: www-math@w3.org
>>Subject: Re: Last Call Working Draft of MathML 2.0, 2nd edition published
>>Resent-Date: Sun, 8 Jun 2003 13:12:41 -0400 (EDT)
>>Resent-From: www-math@w3.org
>>
>>
>>Bill,
>>
>>I am responding to your suggestions about OpenMath URL's.
>>
>>We have elected to go with a slightly modified version of your
>>suggestion, namely using
>>
>>  definitionURL="http://www.openmath.org/cd/setname1#N"/>
>>
>>Due to recent changes on the Openmath site,  the ".ocd" file
>>extension is no longer necessary.  This URL now is valid
>>and addresses identifies a specific dictionary entry.
>>
>>As the exact reference for the BesselJ example is
>>nailed down, I will comment further.
>>
>>Stan Devitt
>>(on behalf of the Math Working Group)
>>
>>    
>>
>
>I am happy with that solution. I would expect that there should be some
>paragraph in the MathML spec. which describes the semantics of this URL
>referance.
>
>Bill.
>
>  
>

Received on Saturday, 28 June 2003 08:13:49 UTC