A Rslt (recursive set of labeled tuples)
is a collection of Exprs (expressions).
Every Expr takes one of four forms.
Phrases can be any string of symbols
dogs, four score and seven years ago, -~~ !*&%
and the empty string are all valid Phrases.
A Rel is a relationship between some sub-Exprs.
An example would be frogs #eat bugs,
which is an eats relationship between frogs and bugs.
Another would be Bill #gave flowers #to Mary,
which is a 3-member gave-to relationship.
Yet another would be #maybe I am immortal,
which is a maybe relationship with the single member I am immortal.
There are two pieces of data needed to define a Rel.
The first are its members --
for instance, frogs and bugs in the first example above.
The second piece of information needed, the Tplt (template),
defines what relationship holds between those members.
Two relationships can have the same members and still be distinct,
so long as their templates are different.
For instance, in the Rel frogs #eat bugs, the Tplt is _ eat _,
whereas in the Rel frogs #(inhabit fewer places than) bugs,
it is _ inhabit fewer places than _.
A Tplt contains a list of Expr values,
which we might call "separators".
In the relationshiop Bill #gave flowers #to Mary,
the Tplt has two separators: gave and to.
A separator can involve multiple words,
in which case it helps to use parentheses,
as in, Bill #(handed or threw) flowers #to Mary.
A convenient way to depict a Tplt
is to intersperse an underscore where a member would go.
For instance, we could refer to the likes Tplt as _ likes _,
and the maybe Tplt as maybe _.
Note that the term "separator" is not exactly accurate:
A separator in a Tplt in Hode usually separates two things,
but in some cases it only has one neighbor.
Examples include the maybe in #maybe I'm amazed,
and the sometimes in everybody #needs somebody #sometimes.
Mary #knows (Bill #likes Sue) is compound:
It is a binary knows relationship between Mary
(which is a phrase, not a relationship)
and the binary relationship Bill #likes Sue.
We say the knows relationship is "higher" than the likes relationship.
(In terms of operator precedence, higher relationships "bind later".)
The Addr (address) a handy way for referring to an Expr
rather than writing it all out.
For short expressions it doesn't matter much,
but for long expressions it can be very helpful.
When an Expr appears in the UI,
its address appears (if the user so chooses) at its left.
That address can be referred to when creating new Exprs,
or when searching for Exprs.
For instance,
suppose giant volcano turtles #lay really hot eggs is stored at Addr 13.
In that case, we could represent
(giant volcano turtles #lay really hot eggs) #during spring
with the shorter expression /addr 13 #during spring.
The definition in Haskell of a Rslt is roughly this:
data Expr =
Phrase String -- word or phrase. (can include spaces).
| Rel [Expr] Expr -- relationship
| Tplt [Expr] -- template for relationships
| Addr Addr -- the address at which some `Expr` can be found
That still omits a couple details; the true definition can be found in Rslt.Types.