mirror of
https://github.com/vitorpamplona/amethyst.git
synced 2026-10-06 19:53:08 +00:00
Vitor PamplonaandClaude Opus 5
288ed01105
`amy sno parse|work|verify` — validate a DECK-0003 payload against §1.9, price the proof of work an avatar owes for it, and check whether a kind:11333 avatar has paid. Local, accountless, thin over quartz: the verbs exist so the reader can be diffed against the implementations the spec points at, without a device or a relay. cli/tests/sno/ is that diff. The reference's own fixtures drive it — _rejections() out of decks/sno-reference.py and the golden vectors out of cyberspace-cli — so the cases are theirs, not ours. 26 §1.9 verdicts agree on both the answer and the rule number; all 8 avatar vectors agree. The four places the two differ are asserted as divergences rather than skipped, because a reader that quietly stopped diverging would be as interesting as one that started: - we read the v:2-with-literal-triples cohort, the arbiter refuses it (§5 permits the generosity; three of seven live objects need it); - we refuse a vertex past 64 units, the arbiter accepts it (§8's own open question, and whole*120 overflows an Int); - the Python reference still calls 33331 the avatar kind, a week behind §8.10's move to 11333, so it reads a conformant avatar as not-an-avatar and would price a standalone object as one; - empty content is §8.10's default avatar to us and not-an-avatar to it. Same outcome — the client draws its default — different word. SnoAvatarEvent gains payment(), which reports required/committed/zeros and why, mirroring verify_avatar_work's shape so the two are directly comparable; isPaid() is now that result's ok. Co-Authored-By: Claude Opus 5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01JwXApJjoZYtkD3sPRWbPNa
2 lines
23 B
Plaintext
2 lines
23 B
Plaintext
state-sno-conformance/
|