ROADMAP opened by saying nothing in it is a promise. Section 7 holds eighteen standing principles cited by number as settled law from commit messages and from other documents - 4 governs what is hashed, 13 defines what a standard is, 17 governs what may write to a record, 18 governs how a contribution is admitted. A reader taking the header at its word would conclude Principle 17 is optional. The disclaimer now says what it always meant: it covers section 4 sequence and section 5 assumptions, not section 7.
Principle 5 gets the same qualification PROCESS section 8 got. A permanent deviation is the specification is right about facts on a host and would be licence to lower a REQ if read without limit.
PRECISION section 10 requires every section 7 entry to say whether it is a limit that work may lift or a boundary that was chosen. The prismatic-only entry said neither, which is the single most consequential ambiguity in the document. It is a limit, and the reason is now recorded: the oracle is a two-dimensional oracle with a scalar length multiplier - every recorded value is a property of the cross-section except LENGTH_MM, VOLUME_MM3 and MASS_G, and volume equals rounded section area times length to within 3.6e-12 across all 113 accepted cases. Lift it carelessly and that relation stops holding; nothing fails, it just stops meaning what it means. Anything keeping each station two-dimensional preserves it.
Also in PRECISION: STL export landed and is no longer planned. An absent export format is a limit; the refusal to emit toolpaths is a boundary.
ROADMAP section 4a corrected to match STAGING-STATE section 3a: nothing here is implemented on Kane Fabric, and SASE was used and never defined. The port is recorded done. The seed commit and test count are removed as derivable.
IDENTITY-CONTRACT: the rewrap owed from a76879f.
Applied by anchored patcher. Suite 643 passed, oracle intact. Documentation only.
ROADMAP.md was described as a reference rather than a commitment -- something
that documents intent and will eventually become a roadmap. The header now says
so, and says that HANDOFF section 4 holds the actual order. Sections 1, 2, 3 and
7 are durable and still correct; section 4's sequence and section 5's production
assumptions have drifted since 18 AUG.
Principle 17: nothing approximate may write to anything reproducible. Principle
4 governs what is hashed; this governs what may reach the thing being hashed. A
search index, a documentation system or a language model may read design records
freely and may suggest anything to a person, but none may originate a parameter
that lands in a record. The failure is silent and compounding -- if suggestions
become records and records become the corpus, the corpus teaches itself its own
guesses, and every record remains perfectly correct about what it was built
from. Where a suggestion informed a build, the record's note field says so,
being already excluded from both hashes. No new field is needed and the
discipline is written down before there is anything to guard against.
Principle 18: a contribution is admitted by the oracle, not parsed by the
compiler. Author in whatever notation suits you, freeze the cases inside the
pinned toolchain, port, and admit when the oracle passes byte-identically. This
has happened twice, with both .scad generators -- it describes what was done
rather than what is planned. Two consequences: no second geometry engine is ever
kept in step with the first, and legacy/ holds the reference tier rather than
superseded code. When someone asks for another input format, the answer is that
notation never runs in production.
A third principle was drafted and dropped -- that artifacts cross a service
boundary and geometry does not. Section 4a's own rule is that use cases precede
interface proliferation, and writing a principle about a kernel service boundary
while the kernel is not a service is precisely that. It is a gate, not a
principle, and the kernel can acquire one when it acquires a boundary.
HANDOFF section 11 gains rows 12 through 16, all found by reading ROADMAP.md
properly rather than grepping it. Section 5 names a production FQDN that the
printed name contradicts. Section 5 says the production proxy is not on the
WireGuard side, for a reason the DNAT at 1fdb115 removed. Section 4a says no use
case crosses the Kane Fabric boundary and that neither project defines the
interface -- membership-gated export is that use case, and IDENTITY-CONTRACT.md
defines the identity half, while the siting interface 4a is actually about
remains undefined and should stay so. ROADMAP section 4 and HANDOFF section 4
are different lists. And legacy/ is misnamed.
None are resolved here. Two are CIVICVS's, two want the whole repository in view
at once, one is cosmetic. Recorded so the reconciliation pass finds them written
down rather than rediscovering them cold.
TheRON