The vendor that I am submitting these XML files to requires entities  
because there are math equations and similar items that can't be  
represented using ASCII. In the layout application, we use a special  
font to generate these items. So whenever they are used within the  
content of the file, we need to change them to the appropriate  
entity. That being said, is there a better way other than manually  
going through my XML file and changing those characters to their  
correct entity code?



chad



On Jul 3, 2006, at 9:21 AM, David Carlisle wrote:



I maintain a DTD that has perhaps more character entities than any  
other

"public" DTD (MathML has more than 2000 entities for characters) but

still, I question whether there is ever a need to have such a  
character

reference in generated files (such as the output of xslt). Entities  
for

characters are really there to ease hand authoring.



If people said they needed attributes to be marked up using ' instead
of " then you would be in a simialr position, the XSLT spec gives no
control over this, so you have the option of

a) educate your users that XML does not distinguish between these  
markup

   forms, so they should not either.



or, if your users are unwilling to be educated, or in a position of
power, such they may choose not to listen, or if you just want to be
kind to them anyway,

b) post process the XSLT output with a NON-XML tool such as perl or  
sed

to edit the markup appropriately.




XSLT1 does give disable-output-escaping that can be used sometimes  
as an

alternative to (b) but it's not usually a good idea, and expplictly an

optional feature in XSLt1 and explictly deprecated in XSLT2. If you

really must do something like this, and can use XSLT2, character maps

provide a much better mechanism architecturally.