I have no desire to become involved in a row over design philosophy or the virtues of existing packages (I personally, despite not particularly liking mice, use xfig to draw my diagrams and export LaTeX picture code), but Paul Taylor's comment (reproduced below) on Ross Street's suggestion is ill conceived:
Suggestions such as Might I also mention strings, links, and Penrose tensor notation. will only cause annoyance. How many such diagrams do you envisage ever being drawn?
My most recent manusript contained fourteen figures, each with between 2 and 9 Penrose tensor diagrams. Joyal/Street's "Geometry of Tensor Calculus I" contains numerous such diagrams as do *many* recent papers on the categories of representations of quantum groups (by such people as Reshetikhin, Turaev, and Lyubasheko). A good syntax for diagrams in Penrose tensor notation, would naturally include a good syntax for drawing knot diagrams (of use to low-dimensional topologist generally) and Feynman diagrams (of use to physicists), besides being invaluable to those of us who work on applications of monoidal category theory in those areas. Obviously, LaTeX is used by many people besides categorists, and if we're going to update it to accomodate our need for algebra freed from the constraints living in linear strings of symbols, we might as well get it right for all people who need such algebra. --David Yetter ==============================================================================