Re: mathml with netscape 7.1 and Mozilla Firefox

David Carlisle said:
>> Hi, you are omiting the DTD for correct sniffing. For the Firefox,
>> Mozilla people also recommends a different namespace declaration in the
>> math element:
>> <?xml version="1.0"?>
>> <?xml-stylesheet type="text/xsl" href="pmathml.xsl"?>
>> <!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.1 plus MathML 2.0//EN"
>>                "" [
>>   <!ENTITY mathml "">
>> ]>
>> <html xmlns="">
>> ...
> It's not usually a good idea to specify the dtd if using pmathml.xsl, it
> will not have any effect on mozilla (unless entity refererences are used
> in the file) but in IE it will cause the browser to fetch the dtd from
> the w3c site which slows things down considerably, and even if the dtd
> is already in the browser cache just reading the dtd can take a
> noticable amount of time.

Well, whereas i could try to understand usage of mathml.xsl on Mozilla i
do not understand usage of pmathml.xsl. Therefore i would simply to
eliminate the second line.

Moreover, a PI element in the prolog for linking external XSLT looks great
for the text/xml or application/xml mimes with xml extension but can
generate problems with a /xhtml mime with .xhtml file.

See further discussion on


About the absence of DTD, it is a non issue. Without the correct DOCTYPE
sniffing, Mozilla will force the quirks mode (non W3C standard) with
different behavior for parsing, scripts, and CSS:


> If you are specifying a DTD, then
> is not the correct DTD to specify, that is an old version. The DTD is
> specified in MathML 2 (2nd ed) as

I was just citing the last DTD recommended by Mozilla developers for the
Firefox. I did not test 'new' DTDs.

> If you are not using pmathml and are allowing mathplayer to auto-detect
> mathml then specifying the doctype is necessary, and doesn't have an
> effect on performance as the dtd is not actually fetched in that case.
> David

Juan R.


Received on Tuesday, 29 August 2006 18:14:11 UTC