RE: brooklyn union gas conjecture Sometimes diagrams are more compact than symbols alone. Both are only as expressive as their inferencing powers! See Etchemandy and Barwise's seminal paper on vvisual reasoning and diagrammatic inference - they rebuke Tennant's position that diagrams can only be syntactic and have no semantic value for proof. Their tool Hyperproof is very OO. Each proof can have multiple representations.
Sometimes the geometric representation of an object is random (ad-hoc) and other times its layout is developed systematically (e.g. the Barker Technique for drawing ER Diagrams, named after the heuristic's creator Richard Barker of Oracle). How would you create a metric for visual vs textual programming "compactness"? At work we have a tool for generating SQL via a visual interface. It takes much less space than the pretty-printed SQL, but you could argue we're using techniques similar to code folding in textual IDEs. Also, what's the "right" pretty printer format to compare against? Some might be bias towards horizontal layout of text, or only horizontally bias for certain grammar constructs (e.g. select-list or select-clause might not line-break for each item in select-list). Also, the tool limits degrees of freedom including nonsensical aggregates such as select c.child_pk, sum(p.some_parent_column) from Parent p inner join Child c on p.parent_pk = c.parent_fk group by c.child_pk Another example would be SQL Server Analysis Services OLAP grid builder. It hides the MDX syntactic cruft through drag-and-drop automated data transformations. _______________________________________________ fonc mailing list [email protected] http://vpri.org/mailman/listinfo/fonc
