Eight libraries, the sqldelight plugin alias and six version keys that no
build script references, and the commented-out lines that named them or
Room 2's aliases. The jvm, debug and release runtime graphs are identical
to Phase 0's.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
marmot-kmp.md's D2 and its out-of-scope item point here; the docs index reads
this note after it; CLAUDE.md says the daemon is Java 21, a first build needs
the network, and a version moves in the catalogue and the lock together. The
plan's status: built but for mls-kmp 1.4.0's release, the mac half of
Phase 6, and Phase 8's manual checks. check green, jvmTest 1392/1392.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
magpieLockAudit runs docs/scripts/magpie-lock-audit.sh from check, beside the
M3, brand, wallet and marmot audits. Every catalogue library in a group Magpie
serves must be declared strictly (D3) and have a lock line for its root,
-jvm and -android modules -- read as text, so a bump made on one platform is
caught on any. Verified both ways: silent on a clean tree; a dropped lock line
or a plain, unlocked bump names the module and fails the build.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
quartz 2.20.0's lines went: under the strict pin, marmot's request for it is
never looked up, and every classpath resolves under a frozen lock without
them. Checked here: a frozen lock refuses a version it does not hold, with
Magpie's own instruction, and an offline build from a fresh daemon resolves
from the mirror. The iOS lines need a mac, and the freeze waits for them.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
giza deleted FoundingAdmins.check (remove-founding-admins-card): it
cannot tell a founder left out by a buggy creator from one the group
demoted, and every founder is an admin when the room is made. RoomAdmins
drops its missing-founders guard, and A4 says why.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
2.20.1 changes two files of quartz's in-process relay server, which neither
the app nor marmot-kmp's main source imports, so it is inert here. The strict
pin wins over marmot 2.3.0's request for 2.20.0, the graph moves by quartz
alone, and check is green: jvmTest 1392/1392, MarmotClientInteropJvmTest 3/3.
mls-kmp 1.4.0 (5b) waits: it is tagged but released nowhere yet.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
C1 measured again on mls-kmp 1.3.1: the old-blob, shipped-profile, store,
sender-ratchet, skipped-generation and fork tests are green untouched, and the
authored fork is still open. The re-save test only said "newer than 4", which
a library that moved to 11 would pass, though a bump is what an older build
cannot read back; the seventh case pins 10 exactly, checked red at 11.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
docs/tools/engine-interop runs marmot-kmp 2.3.0 over mls-kmp 1.3.1 against
quartz 1.15.2's engine -- the one a Bennu peer on that version runs -- in two
JVMs over a pipe, since one build cannot hold two versions of the quartz group.
Each side creates a room in turn; the other joins by Welcome, applies two more
commits and reads a message each way at every epoch. 20 checks, all green, so
mls 1.3.1's new epoch-0 shape costs nothing against the old engine.
The note records Phase 3, including the first draft of the epoch-0 test, which
asserted a restore-time repair mls-kmp drafted and then removed before 1.3.1.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
mls-kmp 1.3.1 gives a new group the RFC 9420 §11 epoch-0 interim transcript
hash, where quartz 1.15.2 and mls 1.3.0 left it empty. A state saved the old
way keeps its empty hash -- 1.3.1's audit commit dropped the restore-time
repair -- so the room the old engine created is restored as it was, and the
commit that adds its first member is the one the hash feeds. Pinned both ways,
and measured: the add, the Welcome, and a message each way across it.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
None of the four is on Maven Central at these versions: they are released on
Nostr + Blossom, and the Magpie settings plugin (0.2.3) serves the quartz,
marmot, mls, crypto, negentropy and schnorr256k1 groups exclusively. The
versions are marmot 2.3.0's own pins (D2), declared `strictly` so a transitive
request cannot move them (D3). gradle/magpie.lock pins the release key and the
manifest of every module resolved -- jvm, android and root so far; the iOS
lines need a mac (Phase 6).
No Kotlin file changed. The resolved graph moved by exactly the four libraries,
negentropy v1.2.3 and the new schnorr256k1-kotlin; secp256k1-kmp still resolves
to the composite. check green, jvmTest 1389/1389 with no test edited, release
minifies through R8.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
The running card names the members whose share it takes away because
the group removed them as admin, and their rows read Share being
rotated away instead of Holds a share. Removing an admin says nobody
is asked again.
The plan is marked built, with where the build chose differently, and
the notes it changes point to it.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Magpie, which the next phase brings in, refuses a daemon below 21. The criteria
file makes Gradle pick 21 itself (foojay downloads one where none is installed),
so a 17 launcher still gets a 21 daemon: checked with JAVA_HOME at 17, jvmTest
1389/1389. No dependency moves in this commit.
The file was gitignored since 8de05742, most likely to keep an IDE-generated copy
out of the tree; it is now the build's statement of its daemon JVM, so it is
tracked.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Every target resolves from the release relay but mls-kmp 1.4.0, which is in
neither Magpie nor Central, so Phase 5b waits and mls stays at the 1.3.1 marmot
2.3.0 was built against. Baselines: 1389 jvm tests in 182 classes, check green.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Make admin at the ceremony's generation no longer sends the admin to
rotate by hand and come back: the offer starts that rotation, which
withholds nobody new and so asks nobody, and the accepted offer's
proposal waits in GovernanceFollowThrough until the group has signed
the new shares. A rotation abandoned or failed ends the offer out loud,
once. The confirmation says the shares are rotated first.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
None of the four is on Maven Central any more: from quartz 2.19.0, mls 1.3.1,
marmot 2.3.0 and crypto 2.1.1 they are released on Nostr + Blossom through the
Magpie settings plugin. The code the app calls has barely moved; the way the
build reaches it has. Nine phases, marmot's own pins first and the newest second.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
A recipient takes part unasked when the rotation withholds nobody the
group has not already decided about: everyone it leaves out until they
are back is stale already, and everyone it leaves out for good is
someone the group's signature took out of a governed room's admins.
The check sits at advance's one gate, so arrival, a replay and the
sweep all meet it, and the room is told in place of being asked.
Every rotation now retires the live holders the group removed as admin
(A3), so whichever runs first after a demotion takes their share away.
Neither applies in an ungoverned room or to a founder the admin policy
has been missing from the start.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
A change that moves who holds a share -- any change of admins, or the
removal of a holder -- is refused while a rotation is under way, until
its key state is signed, and so are offers and step-down requests. A
step-down request or an accepted offer that arrives meanwhile waits and
is taken up on the next pass. A removal whose owed rotation would fail
the majority rule is refused before anyone signs (A12). The member rows
list the rotation as the reason, with whom it waits on.
rotateAway no longer takes a running rotation that still deals to the
removed member for the one it owes, which left them a live share and
said nothing.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
rotationHold gains two reasons: a signed change that moves who holds a
share is still being signed or on its way to a relay, and a new admin's
bound id has no share yet. rotationPending is the part admin changes
wait on, through AdminGovernance.adminChangeHold: a rotation running, or
its key state not yet signed. RoomAdmins reads the live admin policy,
with a seam for the in-memory ritual devices. The Shares section's held
line names the two new reasons.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
A removal's rotation asks every remaining holder. A removal during a
rotation is handed the running one, which deals the removed member a
live share. A new admin bound mid-rotation fails the rotation's round-2
check. A statement being signed fails when the rotation lands. A
step-down request taken up during a rotation is used up. The first
promotion is refused at generation 0.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
FoundingAdmins.check, announceIfMissing, ChatRepository.foundingAdmins and
addAdmins, MarmotOutboundDao.addAdmins and FoundingAdminsJvmTest are deleted:
nothing calls them any more. ceremonyOf, the one piece still used, moves to
MarmotGroupCreation beside uninvitedFounders and finishFoundingInvite.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
The line ended "An admin can add them back from the room's details", and
nothing there does that any more. TYPE_FOUNDING_ADMINS_MISSING stays in
MEMBERSHIP_TYPES, retired, so the rows already written still render as
captions.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
missingFoundingAdmins and addMissingFoundingAdmins go with the card that read
them, and every detail load drops a group decode. founderNames now names only
the uninvited founders.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
The rotation an admin change owes asks nobody, because the group's
signature on the change already decided who is withheld; admin changes
and rotations never overlap. Phase 0's twelve decisions, as recommended.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
A founder missing from a keyed room's policy is now usually one the group
demoted or who stepped down, and the card's "Make them admins" offered to
undo that. The "not every founding member was invited" card stays.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
The retired caption type's place in MEMBERSHIP_TYPES moves to
ChatRoomLastMessagePreviewTest, and the legacy-room case asserts no key
state directly rather than through FoundingAdmins.check.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Since a keyed room can demote an admin and an admin can step down, a founder
missing from the policy is often the group's decision, and the card's repair
offers to undo it.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Brings in a keyed room's step-down by request (bb35d73c) and keyed rooms
born needing the group's signature, with the switch gone (1ccd8c7f).
Resolved: AdminGovernance keeps giza's canAskToStepDown and
stepDownProposer beside the policy-backed memberActions; the
repositories keep memberActionOptions and lose requireSignatures; the
loader no longer asks GovernanceSupport whose devices a room waits on.
Reconciled: the policy offers your own row Step down as admin in a keyed
room on canAskToStepDown's terms -- governed, rotation on, above the
floor, another admin to ask -- so the menu matches the request giza
added. SignaturesNotRequired carries no data and says the group was made
before signed changes and keeps its admins, since nothing switches the
rule on any more.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
An admin's own row had no menu in a keyed room: stepping down was not
offered, because a demotion's proposer starts the rotation that takes
the demoted admin's share away, and a rotation's proposer has to keep a
share.
Step down as admin now sends a request into the room
(PromotionEvents.STEP_DOWN, 30377) and commits nothing. The remaining
admin in the lowest leaf (AdminGovernance.stepDownProposer) -- one
device, named the same everywhere, so two never propose at one epoch --
takes it up in GovernanceFollowThrough and runs the same demote its own
Remove as admin would. A request is used once, and dropped by every
device once its requester is no longer an admin. Offered only above the
floor and while another admin is there to ask.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
The note's status, a table of the build's departures from the plan --
the route carries the action by name, an unavailable item's state rather
than a second click label, the only admin as one reason -- the README
entry, and a pointer from member-admin-promotion's Phase 11. The unused
"See why" string goes with the click label it was for.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Two reasons that clear by themselves, said before anyone tries rather
than in a snackbar after: an earlier change the publish gate is holding
until a relay accepts it, and an admin offer the member has not
answered, by whoever made it. PromotionManager.unansweredOfferTargets
reads the second from the room's offer and acceptance events.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
The overflow menu replaces the chevron on every row but the viewer's
own, whoever is looking. It lists every action that fits the member's
role, open or not, in a fixed order. One that is not open is marked --
the quieter role colour, an info icon, and an "Unavailable" state -- and
opens the screen that says why instead of acting. It is not disabled,
which would leave the explanation unreachable.
Your own row is unchanged: the menu when stepping down is open, the
chevron otherwise. The row says "View profile" to assistive technology
now that the chevron that said it is gone from most rows.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
MemberActionUnavailableRoute carries what was asked -- the room, the
member, the action -- and the screen works the reasons out again
through ChatRepository.memberActionOptions, following the room while it
is open. It leads with the first thing in the way, lists the rest under
it, and offers the one action that helps where there is one: the admin
room for a NIP-17 room, the group's page for the signature switch. A
reason that clears says so.
The action travels by name: the multiplatform navigation-compose has no
NavType for an enum, which the route test caught on its first run.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
The rule that a keyed room's changes need the group's signature was
switched on after founding, once every member's device had said it could
follow signed commits. That protected Bennu members and older Giza
builds from being left behind by the first signed change. Giza is in
development and keeps compatibility with neither, so the switch
protected nobody, and it kept an invited member from ever being made an
admin.
MarmotGroupCreation.createFor now writes GovernanceMarker's line, with
the floor the ceremony's quorum sets, into the room's epoch-0
description. Founding still needs no signature: the founding invite is
the creator's commit while it is the only member, which no other device
processes.
Removed: GovernanceSupport and its event (30372), the announcements on
joining, at start and in answer, the automatic switch and
requireSignatures, and the screen's waiting and ready states.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
ChatRepository.memberActionOptions is the one loader of what each row
offers: the policy over the group's state, with having left and whose
devices a keyed room waits on added from the database. A NIP-17 room and
a legacy room answer from their kind. The view model asks for every
room, not only one this device governs; the row still shows only what
is open until the explanation screen exists.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
MemberActionPolicy decides, over plain sets, every action a member row
offers and what stands in its way: the room's kind, the viewer's role,
the keyed room's two rules, the member's leaf, the floor, and work in
flight. AdminGovernance.memberActions is now that, filtered to what is
open, so its existing tests pin the agreement.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Requiring the group's signature for changes was a button an admin had
to find on the room's screen. Nothing about it was a decision: the two
conditions it waits on -- every member's device can follow, enough of
them are admins -- were already checked in code. The button is gone.
One device switches it on: the admin in the room's lowest leaf, which
in a room as made is its creator (GovernanceSupport.switcherOf). Not
every admin's, because two commits at one epoch are not settled for
their authors (docs/marmot-v2.md, Phase 8); every device reads the same
leaf, so they agree on one without talking. GovernanceFollowThrough
runs the switch whenever a member says it follows signed commits, and
on start.
Not at creation: a Bennu or older-Giza founder would be cut off without
anyone knowing, and a founding invite finished later is a bare commit a
governed room refuses. The room switches seconds after its founders
join instead. The screen names whose device it is waiting on.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>