Minutes June 3rd


Dave Raggett
Patrick Ion
Robert Miner
T. V. Raman   (left early)
Ralph Youngen
Ron Whitney
Bruce Smith
Neil Soiffer

The meeting discussed Bruce's new proposal, see:


Dave asked about lexical details. For instance if one uses an SGML
named character entity how does the tokenizer know whether the
character is allowed as part of an identifier? Bruce replied that
there needs to be a large dictionary that specifies properties such

    o   Whether the character is allowed as the first or subsequent
        characters in an identifier.

    o   If it is an operator, its types (prefix/infix/postfix) and the
        associated left and right precedences.

    o   Whether it can be used to embellish other operators.

Action: Bruce to add a detailed schema for the character dictionary.

Dave also suggested that as a matter of principle any tag names should
have meaningful names. Bruce said he wanted to avoid potential naming
conflicts with other groups wishing to define new HTML tags. Dave said
that this wasn't a big problem, given W3C's role in defining HTML.

Action: Dave to post a proposal for the HTML-math tag names.

Robert added that to allow him to implement the proposal he would need
more detail on the various layout schema. Bruce will work on this.

Dave queried the flat associativity with same precedences for `+' and `-'.
Neil explained that this makes it much easier to write the line breaking

Macro definitions. Bruce will add an SGML element to represent these.
This raises the issue of scoping and how a plug-in could exploit the
HTML parse tree. In the short term, this will remain a problem.

We discussed the representation for arrays. Dave explained that the
HTML 3.0 proposal borrowed from LaTeX and TeX. See:


It supports:

    o   setting position of array relative to
        preceding and followng expressions

    o   column specification for cell alignment

    o   cells spanning multiple rows or columns

    o   "+", "-" or "=" characters as column separators

    o   separation of first row/column as labels

    o   setting left and right bracket symbols

    o   filling a cell spanning several columns with dots

The features needed for math make it inappropriate to use the HTML
table tags.

We discussed what HTML tags might be appropriate within HTML-math.
The current inability to call the browser to handle such nested tags
suggests we need to take a cautious approach. A the minimum we probably

    o   plain text

    o   simple kinds of emphasis (bold/italic)

    o   control over font size

    o   hypertext links

    o   line numbering

We could further allow this text to include math elements so that we get
math including text including math etc. This doesn't seem to be needed
in practice though.

The current plug-in api's are inadequate. For instance one would like to
know the current font family, size and baseline position, as well as the
background color or texture tile and pattern origin. One would like to
set the visible size according to the expression being displayed, and
to be sent a message when relevant parameters are changed. How can CSS
based style sheets influence the style properties used within plugins?
Dave would like the math-erb to put pressure on browser vendors to fix
these problems.

Action: Neil to investigate Netscape Navigator 3.0 plug-in SDK to see
what improvements have been made to the api.

One short term solution would be to add parameters to the math tags
to specify the context in which the elements occur, e.g. <h1>, or <p>.
The control panel for the html-math plug-in would allow the user to
set the font size to be used in these contexts.

We discussed ideas for folding and unfolding expressions. One idea is
to allow the author to name a subexpression and then to use that name
in place of further occurrences of that subexpression. When folded the
given name would be shown in place of the subexpression itself. The
scope for such definitions shouldn't be resticted to a single math
element. This could be supported via SGML tags and attributes.

Bruce talked through the case where names for subexpressions are generated
automatically at browse-time. This doesn't require any special markup,
although the ability to give the same name to common subexpressions will
depend on the ability to recognize that these subexpressions are in fact
semantically identical. In a previous discussion Raman pointed out that
it would be helpful if the user is allowed to set the name of subexpressions
as this makes it easier to remember (important for speech-base browsers).

-- Dave Raggett <dsr@w3.org> tel: +1 (617) 258 5741 fax: +1 (617) 258 5999
   World Wide Web Consortium, 545 Technology Square, Cambridge, MA 02139
   url = http://www.w3.org/People/Raggett