notes.org

tinyurl.com/cas-cockpit

concepts

manipulang is the intermediate representation—basically a lisp, with vectors. terms in operations often represented as binary might need wrapper terms [:sum [:plus “x”]]

fake/real-paths

see L96 in cas.tree_ops

from the mismatch of how [+ 1 2] is both single-level (it’s one vector) and multi-level ( 1 and 2 are children of +)

possibly change operations so up/down etc, make sense with real path?

up: if (last current) !=0, make 0, else subtract last —wrong! see from “+” to 999 in cas.tree_ops, l96 down: add a 0 right: inc last left: dec last

a “real” path is one you can use with what’s in @tree-atom. it leads to “a node,” which may be a vector

there should never be more than one zero at the end of a real path

the first child of a vector should never be a vector, because it is not a “real” child

advantage of “real path”: matches actual datastructure

advantage of fake path: corresponds better to meaning—but does it?

in real path, ending in 0 means it’s an operation, by looking at path alone

in real path, ending NOT in 0 means don’t know whether is vector or value by looking at path alone

however, there is an unambiguous answer to that question, w/tree access

in fake path, no way to refer to first item of vector—it’s the node as a whole

I feel like that’s bad, they should be able to select + and turn it to -

up: selects containing node

down: first child

left: left

right: right

sugar can be written for unintuitive things, but nothing can solve ambiguity or the inability to express

a “fake” path

structure

hotkeys

listeners are like middleware

for tree manipulation, they’re mapped to an “action” (keyword), which dispatches in microsoft_directory_tree

chans

key-chan

comps (board, bench, tree-manipulation)

bench is for one-off components a la devcards

todo

pdf

change backend so a single route can receive a text string, maek pdf, and return pdf