flan/test/programs/values.flan
Joseph Ferano 96ab4c9cf0 Retire the per-type printers, since print says all of it
print-str, print-i64, print-f64, print-bytes, print-line and newline leave
the prelude. print and println are the whole printing surface now, and print
is the better call at every one of the sites that used them: it is the same
structural walk without the newline, so the no-newline case the family was
kept for is covered, and it takes the value as it is. The old print-i64
forced an explicit (i64 x) at every call site, because this language widens
nothing implicitly; that cast is gone from 127 places.

Dropping it moves one answer. hash-grid returns u64, and the cast through
the signed printer showed sand-headless's hash as -2851001042534928384.
print routes a u64 through flan_u64_to_bytes, so it now prints
15595743031174623232 — the same 64 bits, read as the unsigned number they
are. The pinned expectation follows the correction.

test-flan-dev.el and test_session.ml both reached for print-line as "a name
the prelude has"; they reach for rand-seed instead.
2026-09-12 05:32:25 +07:00

22 lines
770 B
Plaintext

;;;; Value semantics, spec-memory.md. Not covered by calc-me, and the property
;;;; most likely to be silently wrong in a backend: a struct or a fixed array
;;;; copies on assignment, a slice copies only its view.
(defstruct P [x i32])
(defvar arr [3 i32])
(defn main [] i32
(let [a (P {:x 1})]
(let [b a] ; a copy, not an alias
(set (.x a) 99)
(print (.x b)) (println ""))) ; 1
(set (at arr 0) 5)
(let [c arr] ; fixed arrays are values too
(set (at arr 0) 77)
(print (at c 0)) (println "")) ; 5
(let [s (bytes "hello")]
(let [v (slice s 1 3)] ; a view into the same bytes
(print v) (println ""))) ; el
0)