Repository navigation
test(messaging): cover the webex spool notice against the real egress gate - #10362
Conversation
Design Review (Fable 5) — ✅ PASSDesign-level review of Design-Verdict: PASS A narrow, real gap (the room-id arm had zero spool coverage) pinned through the real Suggestions
[DESIGN-REVIEWED] 19fc4c7 |
First Principles Review (Fable 5) — ⏭️ skippedRevision |
GPT 5.6 Review — ✅ no blocking findingsGPT 5.6 completed its review of This comment is updated in place on each push. Review detailsNo findings. False positive or not applicable? A repository writer can comment: |
Opus 4.8 Review — ✅ no blocking findingsReviewed Verdict parsed from the review's SHA-scoped output markers for commit False positive or not applicable? A repository writer can comment: |
8386d2e to
f414d71
Compare
f414d71 to
74f2381
Compare
74f2381 to
7d45ac7
Compare
910f477 to
f3c64d8
Compare
c18c097 to
4ddf4e7
Compare
|
@cixuuz update: the adoption half of this PR was overtaken by #10047, which landed the inbound spool on webex/wecom/imessage (and the rest) on main, so I've reset the branch to main and reduced this PR to just the Webex egress coverage you asked for. Main had none: no webex mention in Two tests against the real |
4ddf4e7 to
e662031
Compare
|
Withdrawing the revoked-space test on Design Review's finding, which is correct. A room-root webex space route does carry the sender as principal: the dispatch writes This PR is reduced to the one true test, |
… gate One regression pinning that WebexTransport.may_send_to grants a spooled space route carrying no principal via its room-id arm, so the restart notice posts at the room root. Main had no webex spool coverage. The deny side is not pinned because a room-root space route does carry a principal today, so that guarantee is not true and the behaviour is with a maintainer. The adoption half of this PR was superseded by #10047. Refs #8912
e662031 to
19fc4c7
Compare
iamwhatever
left a comment
There was a problem hiding this comment.
Test-only addition of one regression over the REAL Webex egress gate: I verified in src/kiro_crew/webex/transport.py that may_send_to checks the room-id arm (_allow_group_rooms and conversation_id in _allowed_rooms) before the principal arm, so the test's premise holds and only send_message is stubbed. asyncio, Any, spool_home and _spool all pre-exist in the file. No production code touched, all checks green (the 6 CANCELLED entries are superseded duplicates of runs that later succeeded).
… gate (kirodotdev#10362) One regression pinning that WebexTransport.may_send_to grants a spooled space route carrying no principal via its room-id arm, so the restart notice posts at the room root. Main had no webex spool coverage. The deny side is not pinned because a room-root space route does carry a principal today, so that guarantee is not true and the behaviour is with a maintainer. The adoption half of this PR was superseded by kirodotdev#10047. Refs kirodotdev#8912
What is the problem?
WebexTransport.may_send_tochecks the room-id arm first (room_id in _allowed_roomswith group rooms on) and grants there, then falls through toprincipal in _allowed. So a space route with no principal is authorized by itsroom_idalone. Main has no test for this path:test/test_inbound_spool.pynever mentions webex, and no spool test sets_allowed_rooms.What this PR is
One regression over the real Webex egress gate for a spooled restart notice:
test_a_webex_allowed_space_gets_its_notice_at_the_room_root-- an allow-listed space (_allowed = frozenset(),_allowed_rooms = {"SPACE9"}) is delivered at the room root (sent == [("SPACE9", None)]), proving the room-id arm grants a space route that carries no principal.It uses
WebexTransport,WEBEX_CAPABILITIESand the sharedreplay_spooledonly, and the docstring states exactly what the asserts prove and no more.What was superseded
This PR originally adopted the inbound spool on webex, wecom and imessage. #10047 landed that adoption on main for all remaining channels, so the dispatch-site changes and the spec edit are dropped -- main owns them. What survives is one piece of Webex egress coverage main still lacks.
What tests we did
timeout 900 python3 -m pytest -n0 test/test_inbound_spool.py -x -qagainst current main: 72 passed (main's 71 plus this one). black, isort, flake8 clean; comment-history gate clean.Refs #8912 (satisfied by #10047, not by this PR).