-- Kebab-case identifiers vs binary subtraction.
-- A dash inside a name is part of the identifier; the lexer is greedy
-- (`[a-z][a-z0-9]*(-[a-z0-9]+)*`), so `best-d` is always one token, never
-- `best - d`. To subtract, separate the operator with spaces.
-- Binary subtraction: prefix `-` followed by two atoms, spaces required.
sub-explicit best:n d:n>n;- best d
-- Kebab-case lookup: `best-d` is a single identifier, distinct from `best` and `d`.
look best:n d:n best-d:n>n;best-d
-- All three coexist: same scope binds `best`, `d`, and `best-d`. Each
-- resolves independently. `str best-d` prints the kebab value (99),
-- proving the parser never splits the ident into `best - d`.
mix>L t;best=10;d=3;best-d=99;[str best-d, str best, str d]
-- Multi-segment kebab works the same way.
chain>n;a-b-c=42;a-b-c
-- Subtraction between two hyphenated names: the spaces are load-bearing.
-- The mandelbrot pattern `zr-sq - zi-sq` works because the lexer sees
-- three tokens (Ident `zr-sq`, Minus, Ident `zi-sq`). Without spaces it
-- would collapse to one 4-segment ident `zr-sq-zi-sq` and ILO-T004 would
-- fire with a hint pointing at the bound `zr-sq` / `zi-sq` split.
mandel zr-sq:n zi-sq:n>n;- zr-sq zi-sq
-- run: sub-explicit 10 3
-- out: 7
-- run: look 10 3 99
-- out: 99
-- run: mix
-- out: [99, 10, 3]
-- run: chain
-- out: 42
-- run: mandel 25 9
-- out: 16