The result of the bullpen campaign, carried out by @deus as operator with
@coder on versions.lua (job 1158). Both conformance suites green, attested with
0 open points; the record is on branch kampagne/2026-07-28 in this repo.
What it does:
versions.lua M.SUPPORTED gains 2026-07-28, old version first so an existing
client keeps negotiating what it always did.
lmcp.lua requires envelope; routes on req._legacy, so initialize, ping
and logging/setLevel keep answering the old shape while the new
protocol gets the envelope path. resultType defaults to
"complete" on every result.
envelope.lua new -- header and body checks, returns the -32020 family.
104 of the 180 added lines in lmcp.lua are comments: the campaign also carried
the German-to-English translation, so the diff looks larger than the behaviour
change is.
NOT yet on master. lmcp runs on around ten hosts in this fleet; a green suite
says the spec is satisfiable, not that it is complete or true. This branch
exists so a second pair of eyes reads it before it reaches any of them.
The copy fix in versions.lua (err.data.supported handed out as a COPY, so a
caller cannot hijack the module table) predates this work -- it was the
uncommitted German state, archived on wip/lmcp-deutsch-2026-08-09.
Phase B, third attempt. The contract is the head of tests/phase_b_acceptance.lua;
this is the module that satisfies it.
M.SUPPORTED = {"2025-06-18"} -- and that list is not a guess. The acceptance test
anchors it against a LIVE server: it calls initialize and compares M.SUPPORTED to
the protocolVersion the server actually reports. The first attempt shipped
{"2025-06-18","2025-11-25"} and passed 14/14, because the test then only compared
the list against itself; lmcp does not speak 2025-11-25. Answering HTTP 200 to a
version header proves nothing -- the same endpoint answers 200 to 1999-01-01.
Provenance, all of it in marfrit/bullpen room.jsonl:
#218/#234 @coder wrote the module (qwen3.6-coding), two passes rejected first:
pass 1 implemented M.is_supported instead of the contract's M.check
and embedded its own self-test; pass 2 arrived without a certificate.
#246 @testdesigner refused to materialise the file in the clone root --
writing the subject and then greening it yourself collapses the
implementer/tester split. The refusal was correct.
#256 two certificates, both from container `testdesign`, same test
(sha256 78a4758d706a2539, commit e8c18f1e3b):
empty state -> ROT rc 2 (subject absent)
these bytes -> GRUEN rc 0, 17/17, live anchor included
#283 @deus lifted it out through the room; sha256 byte-identical.
sha256 of this file: e5bb0739b84a2354422842cc1be28d2d8ad407307b84e3f05e79faa30e68e5da
marfrit
marfrit