From: <osserman@math.mit.edu>

Date: Sat, 28 Jun 2003 00:54:39 -0400

Message-Id: <200306280454.h5S4sdM11123@lagrange.mit.edu>

To: www-math@w3.org

Date: Sat, 28 Jun 2003 00:54:39 -0400

Message-Id: <200306280454.h5S4sdM11123@lagrange.mit.edu>

To: www-math@w3.org

Hello, I have been looking at the MathML spec, and I had some questions regarding intentions for labeled diagrams and figures, both of which arise frequently in my field (algebraic geometry). I apologize if these issues have come up before, or if I missed something in the spec, but I couldn't find answers to these questions either in the spec or the list archive. As far as I have been able to tell, the current intent for, say, commutative diagrams, is to offer limited support via mtable. However, I am not clear on exactly what the capabilities in this regard would be. It seems that one could do unlabeled diagrams, even with diagonal arrows, but that labeling the arrows (other than horizontal ones) might not be possible. Is this correct? What is the rationale for not having some degree of explicit labeled diagram, perhaps along the lines of what the xymatrix package offers in latex? Along the same lines, I was looking through the list of arrows available, and I was able to find most of the simplest ones that would be likely to come up in diagrams, but I couldn't find two fairly common ones: dotted and two-headed (ie, for surjective maps) diagonal arrows. These seem to me to be far more basic than a number of the arrow symbols which are included; am I just missing them somewhere? Finally, with regard to images, I see that you have decided to omit any specific support for them. While I can appreciate your attitude that content such as labeled diagrams would work better on a philosophical level as a structured graphics/label encoded entity than an image, and I wholeheartedly agree that such a construction is desireable, given that you have decided to not include such a construction, I'm not sure I see the wisdom in choosing to not support the obvious fallback construction, which is to say diagrams as images. If there is no general labeled diagram facility, and no support for images for "complicated commutative diagrams", what do you recommend that authors do when they have such a diagram? I also take issue with the assertion that glyphs and labeled diagrams are the primary situations where one would want to use an image (at least, presuming that 'glyphs' is in reference to representations of characters and atomic symbols; I'm not sure what the standard usage is). The most obvious alternative is figures, which are widely used throughout geometry, topology, and calculus, and which I believe semantically make more sense than labeled diagrams to be encoded as images. It seems to me that simple support for images with MathML tags placed at specified locations (similar to xfig's support for figures encoded in a combined ps/latex format) would give the semantically and practically desireable implementation for figures, and would also provide a very workable backup plan for labelled diagram implementation until something more robust can be adopted in the future. I would be very interested to hear what the working group's perspective is on these issues. Thank you for your assistance, BrianReceived on Saturday, 28 June 2003 01:13:47 GMT

*
This archive was generated by hypermail 2.2.0+W3C-0.50
: Saturday, 20 February 2010 06:12:55 GMT
*