- From: Neil Soiffer <Neils@dessci.com>
- Date: Sun, 30 Mar 2008 14:22:55 -0700
- To: "William F Hammond" <hammond@csc.albany.edu>
- Cc: public-html@w3.org, www-math@w3.org
- Message-ID: <d98bce170803301422y1669bb1fx829eb60b2b7ff584@mail.gmail.com>

Many (most?) of the WYSIWYG editors make use of the semantics tag. This list includes MathType, Mathematica, Maple, and OpenMath. Disallowing it will mean that most MathML that currently exists or will exist for the next few years will not be valid HTML5. I hope that compatibility is a priority for HTML5. There seem to be two arguments against semantics: 1. the contents can be out of sync with the presentation 2. It lessens the incentive for developers to polish MathML import MathML has been around for 10 years. Although I'm sure that contents can and have gotten out of sync, I have never heard anyone complain about that -- 10 years of experience with MathML shows that this theoretical problem not a problem in practice. Does anyone arguing point 1 have some data to back up their claims that it really is a problem? The second point could be rephrased, "it makes it easier for developers to round trip to/from MathML because they can have exact fidelity and preserve options that may not be part of MathML." As a theoretical example, suppose some fancy math typesetting system has a way to specify the angle of the bevel in a beveled fraction. MathML doesn't contain support for that. They may also have support for how much overhang a fraction line has, or have control over various aspects of drawing a radical, etc, all of which are not part of MathML. There are several semi-kludgy methods for embedding extra info in MathML (eg, using namespaced attributes), but the ability to simply include their format means that they can safely embed their data quickly and easily and still support MathML import/export. Does anyone arguing point 2 have some data to show that vendors that use semantics did that in lieu of polished import? Neil Soiffer Senior Scientist Design Science, Inc. www.dessci.com ~ Makers of Equation Editor, MathType, MathPlayer and MathFlow ~ On Sun, Mar 30, 2008 at 12:58 PM, William F Hammond <hammond@csc.albany.edu> wrote: > > David Carlisle <davidc@nag.co.uk> writes: > > > It's not that that is not a valid concern. ideally you want the rule > > that if any branch of a semantics element is edited, all other branches > > are regenerated (or discarded if that isn't possible) but to simply say > > that sematics shouldn't be supported would be a very backward step. > > But part of the idea with html5, illusory or not, is direct authoring. > Automatic regeneration is beyond the pale in that context. > > > So the preferred solution is to use presentation mathml on the web page > > and annotate it with the content mathml. To just serve the presentation > > form over the web reduces the semantic content of the document, harms > > accessibility and harms re-use of the expressions. Of course simple > > cases you can re-process the presentation mathml and infer the > > semantics, but in many cases you can not, which is the whole point of > > content mathml. > > Is it unreasonable for me to think that semantic seriousness should > live only on the xhtml side, especially if the html5 folk want to bag > semantic annotation? > > (I sense semantic seriousness in only a small minority of all math > authors.) > > > -- Bill > >

Received on Sunday, 30 March 2008 21:23:30 UTC