================================================================================
P9610   T.B. Dinesh & S.M. Uskudarli
        "Specifying input and output of visual languages"


We consider ``context free'' specification of visual syntax -- with the aid
of visual (i.e., 2-dimensional and user-defined) ``lexicals''. We argue
that context-free visual syntax specification can go a long way towards
specification of visual languages in the presence of certain intelligent
tools. We focus on the input and output behavior  of visual languages in
demonstrating this point.

The usual notion of term construction in a editor is extended by allowing
sharing of sub-terms. We demonstrate the utility of this for constructing
input terms (programs) of visual languages. We also allow such sharing of
sub-terms in equations editors which are used to specify language
semantics. This sharing not only allows the building of appropriate
(visual) terms but also provides information on the desired view of an
output term. We introduce a technique called ``Share-Where'' which is used
to preserve layout information and demonstrate its use as well as its
limitations.

We then describe an extension to our formalism which allows dynamic input
and output -- interaction during term execution.  We demonstrate this with
a full example for the FSA language to animate FSA term evaluation.