Re: translating 12083 to HTML-Math layout schema

[was: Re: notes on conference call of 15 July]

At 8:23 PM 7/15/96, Ron Whitney wrote:
>Bruce, on response to the notes of 15 July:
>>> Can you clarify the suggestion? I'm not completely sure what is being asked.
>Perhaps vaguely, I have the idea that layout schemata such as
>mfraction, mroot, mscripts, munderscript, moverscript, mprescripts,
>etc.  correspond to ISO 12083 elements fraction, radical, sup, inf,
>top, bottom, etc. (the correspondence isn't quite parallel as given
>here).  Generally, my understanding of the layout schemata is that
>they are our abstract handle on visual layout:
>   Every expression in HTML-Math ultimately specifies the relative
>   sizes and arrangement of a collection of symbols layed out in a
>   "logically 2-dimensional" manner. This structure is specified not
>   as coordinates, but in terms of a small set of "perceptual
>   primitives" or "layout schemas" which are sufficient to describe
>   almost all of the commonly used notations in existing typeset
>   mathematics. This choice of level of representation is both as
>   general and as abstract as possible while still being based on
>   the structure of the notation for an expression, rather than
>   purely on its semantic structure or meaning.
>                                                - Wolfram proposal
>The "logical 2-dimensional layout" is also what ISO 12083 (the current
>version, not the update being generated now) addressed.  These weren't
>the words used by the standard, but I think that was the spirit.  So
>my question is whether the structure of the ISO standard and our
>method of logical 2-dimensional layout can be aligned so as to make it
>possible for people to filter directly (and without hand-standing) to
>the display list format and to avoid the expression-tree (which
>resides at a logically higher level in my view, although Neil was
>reluctant to agree to this).  I'm just thinking that this creates a
>smooth connection between our work and the world of 2-D layout.
>Upgrade to expression-level wouldn't be required, but since we will be
>supporting a layout level, we support the work of others if we make
>our layout primitives equivalently "universal" as those of 12083.  The
>"logical" 2-D layout model is clean insofar as it is well-understood
>and "closed" in certain natural ways.
>I'm not sure this answers your question, though, Bruce.  Perhaps I'm
>still waving my hands too much.  Let me know.

Not at all, this is quite clear. I agree with you that the expression
tree is, in general, at a logically higher level than the display list.

Note that our proposal permits the display list to be directly
specified when desired, using special markup for each layout schema.

(In principle, transformation rules can modify that, but the built-
in ones won't, and authors needn't add rules that do if they want to
do what you're suggesting.)

So the only issue is making sure we are at least as universal as 12083,
and that there's a reasonable way to translate from it to us.
(And I suppose there is some application for translating the other way
as well.)

Unfortunately I don't know 12083, and from what I'd heard,
I thought its layout was not based on a general "nested expression
structure" like we're discussing. Can someone who knows it well
enough comment on this?