RE: Where do the "default renderings" in spec chapter 4 come from?


Thomas' position below is more or less my view too.  A suggested rendering in a world of many languages and notational conventions is inherently inadequate, and that to do better is going to involve custom XSL transforms or something like that.  At a conceptual level, any system that seeks to display mathematics to readers using the conventions most familiar to them has to start from an abstract semantic representation and transform it to a presentational representation based on the profile of the reader.  

MathML has gone farther than most to provide the basic ingredients in this picture in the form of presentation and content MathML, I think, but it is still a challenging task.  Over the years there has been much discussion of whether MathML itself should try to provide the transformation machinery itself, and there are some arguments to be made for that.  But at present, using standard technologies like XSLT is still the way to go.

As far as where the default rendering came from, what David has previously written is true, and lots of people have worked on them and made suggestions over the years.  However, I think that I am probably originally responsible for quite a few of them, which accounts for the bias towards notations popular in the US.

Seems like the implication of this discussion is that we should try harder to make it clear in the spec that the default renderings are sample renderings, and that doing a really good job of presentation requires something like the picture outlined above.


Dr. Robert Miner
W3C Math WG co-chair
Vice President, Research and Development
Design Science, Inc.
140 Pine Avenue, 4th Floor
Long Beach, California 90802
Main:   (562) 432-2920
Direct: (651) 223-2883
Fax:    (651) 292-0014

> -----Original Message-----
> From: [] On
> Behalf Of Thomas E. Leathrum
> Sent: Tuesday, January 13, 2009 11:11 AM
> To:
> Subject: Re: Where do the "default renderings" in spec chapter 4 come
> from?
> It seems to me that the whole point of implementing the suggested
> rendering in an XSLT stylesheet (rather than trying to enforce
> something
> using DTD or Schema somehow) is that it is non-normative -- local
> systems can, will, and should implement their own renderings by
> modifying ctop.xsl.  I know I have made extensive changes to my local
> copy of ctop.xsl (for Content MathML2), and I know others who have done
> the same.  Saying that the W3C suggestions have a large impact is no
> more universal than saying that US notation conventions have a large
> impact (in most cases, the ctop.xsl default renderings seem to follow
> US
> conventions, for better or worse).  In my experience, it has been well
> worth the effort of learning enough XSLT to make such changes.
> <link to YouTube video of Leathrum climbing onto soapbox....>
> On a tangential note, this is actually one reason why I don't much like
> the idea of a CSS profile for Presentation MathML -- I understand the
> practical need for it, but stylesheets are supposed to be non-normative
> and localizable, whereas Presentation MathML is supposed to be
> normative
> and static.
> <link to YouTube video of Leathrum climbing off of soapbox....>
> Regards,
> Tom Leathrum
> Christoph Lange wrote:
> > On Tue, Jan 13, 2009 at 1:28 PM, David Carlisle <>
> wrote:
> >
> >>> Where do these default renderings come from?
> >>>
> >> So the answer to the question in the case of mathml2 is that the
> choice
> >> of notation is at the editorial discretion of whoever was editing
> the
> >> file at the time, in consultation with the working group members (at
> >> that time).
> >>
> >> … the notation given is probably, in the mind of a
> >> probable editor, the most commonly used notation in the country in
> which
> >> the probable editor was probably educated.
> >>
> >
> > Thanks, David, that answers my question.  Actually, I intended to ask
> > about this process of agreement, not about the technical generation
> of
> > the renderings.  Has there ever been a debate on a particular
> > rendering, e.g. that editor A suggested one rendering that editor B
> > didn't like?
> >
> >
> >> Which is why the spec has alwys been explict that these are
> essentially
> >> only suggested defaults and that a system may (and systems do) use
> >> notations.
> >>
> >
> > Sure -- but given the importance of MathML, your "suggestions" do
> have
> > a high impact.
> >
> > Cheers,
> >
> > Christoph
> >
> >

Received on Tuesday, 13 January 2009 18:03:36 UTC