W3C home > Mailing lists > Public > www-math@w3.org > July 2002

Re: MathML Test Files are cheating!

From: Robert Miner <RobertM@dessci.com>
Date: Wed, 31 Jul 2002 09:45:52 -0500
Message-Id: <200207311445.JAA03332@wisdom.geomtech.com>
To: piret.stephane@ibelgique.com
CC: www-math@w3.org, www-math@w3.org


> I've had a look at the MathML Test Files. I think
> some tests are cheating!
> The source given sometimes refers to entities like
> &ExponentialE; , &gamma; , &infinity; ,  ect..., but
> the actual source uses codes like &#x02147; ,
> &#x003B3; or  &#x0221E; .

Depends on your definition of cheating.  Perhaps there should be a
link to a note from each test explaining what's going on.

The numerical values are the result of expanding the entity names
according to the entity declarations in the MathML DTD.  The reason
this is done is that the MathML DTD must be explicitly referenced if
entity names are used.  That makes the tests intolerably slow on some

The reluctant consensus in the MathML/XML community seems to be that
entity names are not really practical in many/most settings, and
therefore either raw UTF8 characters or numerical entity references
are preferred.  However, it is clear that entity names are preferable
for human readablity, so that is what we used in the 'quoted MathML'
shown on each test page.

The testsuite does, however, contain tests specifically for entity
names.  In the "Characters" section of the testsuite, you will find
tests in which character are encoded using a) entity names, b)
numerical references, and c) raw UTF-8.


Robert Miner                                    RobertM@dessci.com
MathML 2.0 Specification Co-editor                    651-223-2883
Design Science, Inc.   "How Science Communicates"   www.dessci.com
Received on Wednesday, 31 July 2002 10:46:51 UTC

This archive was generated by hypermail 2.3.1 : Tuesday, 6 January 2015 21:27:32 UTC