TODO.md

In decisionpatterns/operator.tools: Utilities for Working with R's Operators

• Resolve name convention. Is it inverse, invert, complement? How does this work with formula.tools and venn?

• There is some confusion about the SPECIAL CLASS of operators, ie. registered, unregistered, etc., and it's type, i.e. logical arithmetic ... we should factor this out and call it: domain

• Relational operators are also logical. operators might have more than one type and have an inheritance behavior

• Create 'function.tools' package

• extend as.name, as.function, as.character, as.operator, etc. to replace fun2name, name2fun, etc.

• create options('functions') to contain metadata for functions

• THEN:

• create 'operator.tools::operators' classes that extends function

• refactor so functions can work on collection objects/lists

• extend so that %in% works for class name > as.character( op(x) ) %in% c( '==', '%in%' ) [1] TRUE > x Species %in% c("setosa", "versicolor", "virginica") > as.character( op(x) ) %in% c( '==', '%in%' ) [1] TRUE > op(x) %in% c( '==', '%in%' ) Error in match(x, table, nomatch = 0L) : 'match' requires vector arguments > class(op(x)) [1] "name"

• There seems to be some friction with rel.type function and expressions this could work better.

? Can operators have multiple types? This might happen with overloading since operators are special functions with two or more arguments.

? setInverse for adding to altering .Options$operators.inverse

------------------------------------------------------

COMPLETED:

------------------------------------------------------

x fun2name: given the function, it identifies the .name function (e1, e2) .Primitive(">=") >===> >=

decisionpatterns/operator.tools documentation built on Sept. 21, 2017, 10:30 p.m.