[cmj0121]

Closes #360.

What

Two tuples, or two arrays, order with < <= > >= by their parts, as docs/core/types.md already said:
the first part that differs decides, and nothing after it is read.

print (1, 2) < (1, 3)            # true
print ("apple", 2) < ("banana", 1) # true
print (1, 2.5) < (1, 3.5)        # E3044 … and a float has no `Ord`

Decisions made here

• Built, not marked [not yet]. The issue allowed either; the == half was built in two containers compare with == and != #295 with the same
  binding and holding, so the ordering reuses that lowering and only the test differs.
• The four operators come from one less, as Ord's do (a <= b is not (b < a)), so each operand is
  evaluated once. A part is asked < both ways to tell "differs" without asking == too.
• What orders as a part: an integer of any width, a str, a type with an Ord, or a tuple / array whose
  parts do. A float does not (a NaN breaks a total order, which is why it sits out Ord), and neither does a
  bool.
• A part without an order is refused for that part, under E3044 — the same rule as two operands that are
  not numbers, so one report site with a different subject (error-codes-check holds one site per code).
• Lists, maps and sets still have no order. The chapter names a tuple and an array only.

Verified

• Local gates, one at a time, each read, after the last change: build fixpoint lint fmt-self error-codes-check
  seed-gaps refuse reject mem-check oracle corpus sanitize-corpus mem-peak examples docs-mirror docs-links
  docs-zerg docs-refused docs-repeat marker-codes chapter-codes deviation-check refusal-cites method-gaps gates
  treesitter dead-code stdlib-test compiler-test check-equal reject-fuzz fmt-corpus behaviour conformance
  productions suites desugar. reject and mem-check also under /bin/bash.
• The new mem-check case was red with the left operand's give-back removed for an ordering (780 per round),
  green after.

Found on the way, not fixed

• docs/core/specs.md says impl Ord for P is E3013 no spec named Ord; the corpus case writes
  impl Ord for V and it builds and dispatches. That [not yet] looks stale for Ord.
• sort or a T: Ord bound over a tuple was not measured.

Not verified

• make test as one board was not run.

🤖 Generated with Claude Code

https://claude.ai/code/session_01N8jd1oK2orM6diHxhvWhC7