Pixelfed 0.14.4 (FrankenPHP with Horizon and the scheduler, on the shared
Postgres and Redis) and WordPress 6 with ActivityPub 9.3.1 (on a new shared
MySQL, WP-Cron run by a sidecar) are peers now, each with its scenario:
Pixelfed 26 checks (photos both ways with alt text, its place arriving as
Rome with its coordinates, comments, likes, a boost, edits and deletes),
WordPress 17 (an Article with its title, our reply, like and boost kept as
comments of their kinds, its edit and removal). Pixelfed has a town driver
and a pair spec: 145 checks pass, 2 expected (G-0007).
What they showed: Pixelfed names our posts by their page (fixed in
26dac40); on PostgreSQL its migration making caption nullable never runs,
so every remote boost failed (the pasture applies it); it files a DM it
fetches as followers-only (G-0007, upstream); Passport refuses a token
whose user id equals its client's id. WordPress signs with RFC 9421
first, which PrivaPub now verifies (c5a69d2).
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01LsXgEaXee4GCU1hwYgPJXw
Lemmy's moderation reached PrivaPub only as removals. Now a remote
community's lock and ban, relayed in its Announce, apply too:
- a lock (Announce{Lock}, or commentsEnabled false on the post) refuses
replies to the thread, ours included, until Undo{Lock}; statuses say so
in privapub.locked;
- a ban of a persona (Announce{Block} with the community as target, or the
moderator's own Block sent straight to us, which is the community's ban
and never the moderator's block of the persona) shows as blocked_by on
the community and refuses the persona's posts and replies there until
the Undo.
The Lemmy scenario's removal was an expected failure only because it gave
up before Lemmy's 30-second batch; it now waits, and checks the lock and
the ban live (Lemmy refuses a lock or an unban without a reason): 29
checks, none expected to fail. G-0003 and G-0006 are closed.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01LsXgEaXee4GCU1hwYgPJXw
A comment in a community was announced only as Announce{Create}, which
Akkoma, Mastodon and Misskey drop: their members never saw it (the village
found Akkoma's missing). A comment is now also announced as itself, as a
new post already was; Lemmy answers that form 400, harmlessly. Nothing is
decided by the follower's software.
The town's checker expects a followers-only quote to stay with the
followers, and G-0003 covers only Lemmy-hosted communities (their relayed
moderation), since relayed likes count.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01LsXgEaXee4GCU1hwYgPJXw
PeerTube 8.3.1 runs on the shared Postgres and Redis, transcoding off. Its
scenario passes 24 checks: a persona follows a channel; a new video comes as
the channel's Announce and reaches the persona's home as a boost, playable
through the media proxy with byte ranges; comments both ways thread; a like
and its undo count; renaming and deletion arrive; the unfollow; statistics.
The town's seeder also takes reruns on the same accounts in its stride: a
Lemmy community already made is found, a circle member already approved
asks nothing again.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01LsXgEaXee4GCU1hwYgPJXw
decePub's e2e boost on Misskey undoes its boost and finds the count still
up: Misskey deletes the renote but its NoteDeleteService lowers only
repliesCount. Recorded as a peer gap, so the cell is a known failure.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01LsXgEaXee4GCU1hwYgPJXw
Pleroma 2.10.2 runs in the pasture from its OTP release (libvips, the
dedupe question, and hackney pointed at the system CA bundle); its driver is
Akkoma's on its own container and database. The pair passes 241 cells, after
its first run showed Pleroma dropping Follows it could not verify yet.
The PrivaPub driver finds the group an earlier run made instead of failing
on the name; the seeder posts a quote or reply of a post from its own round
after that post.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01LsXgEaXee4GCU1hwYgPJXw
Iceshrimp.NET 2026.1.2-beta runs in the pasture with AuthorizedFetch on;
its driver registers through the native API and takes Mastodon tokens from
its OAuth form. The pair passes 225 cells. A new Iceshrimp note reaches its
author's followers only, never an account it mentions or answers, until it
is edited (G-0004, peer): delivery cells now say when a server holds a post
only because it was addressed.
The checker credits a seeder's fetch to the object fetched, not to the reply
that needed it; the seeder skips the wait for an accept whose follow was
already accepted; the PrivaPub driver signs its roots in again when a run
reuses saved sessions.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01LsXgEaXee4GCU1hwYgPJXw
The report merges decePub's end-to-end cells (out/client.jsonl, the
latest result of each test) into the matrix under the observer
"decepub", and names its run without the workstation's paths. A second
spec, hollo-pair, seeds PrivaPub and Hollo alone (258 of 261 checks pass;
the three left found the emoji variation selector bug fixed in
PrivaPub), and Hollo accounts may join circles. The seeder waits up to
90 s for a follow request, two of a sender's retries after a refused
first delivery (Hollo answers its first concurrent deliveries 500 while
it creates its tables). town.sh drive prints its JSON unescaped.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01LsXgEaXee4GCU1hwYgPJXw
tools/pasture/town/ (run through tools/pasture/town.sh) seeds a fake
community across every running peer and checks that all of them, and
PrivaPub, agree on what happened:
- drivers per platform on four dialect bases (Mastodon API, Misskey API,
Lemmy API, PrivaPub with /clientapi), each with a selftest against
its own server; what a server holds is read from its database, never
by making it fetch;
- a deterministic generator (specs/village.json: 23 accounts on seven
servers, roots with several personas, circles and communities, a
cross-server follow graph, posts of every kind and visibility, reply
rounds, likes, boosts, reactions, votes, edits, deletes, blocks,
mutes and a report) and a seeder that keeps a ledger of what happened;
- a sweep that expects delivery and confinement per server, what each
account sees, counts, threads, edits, deletes, follows and privacy
rows (sibling keys, published days, canary root credentials in every
peer's database, located posts that never leave), with what the peers
do on purpose modelled (Misskey drops orphan replies, Lemmy keeps only
community content, edits go to the post's own audience);
- known gaps (gaps.json) turn failures into xfail and passes into xpass;
a self-contained report.html, and docs/INTEROP-BACKLOG.md.
The pasture moves to a public-looking subnet (peers with no private
address switch can join), takes PASTURE_PORT when 6971 is in use, adds
peers to a running pasture (run.sh add, Caddy recreated with its CA
kept), removes its volumes on down, writes every scenario check to
out/scenarios.jsonl, serves decePub as decepub.test for its e2e tests,
lifts GoToSocial's and Lemmy's own rate limits, trusts Caddy in
Mastodon (TRUSTED_PROXY_IP) and gains Hollo (Fedify), whose one login
owning several accounts is the nearest peer to PrivaPub's personas.
The first village found the four PrivaPub bugs fixed in the commits
before this one; the second run, on the fixed server, passes 2319 checks
with 11 failures left, all between peers or from Lemmy's send worker,
which the seeder now warms up first. ROADMAP records the owner's
decisions of 2026-10-04 (the town, and P9 back from the cut list).
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01LsXgEaXee4GCU1hwYgPJXw
thepraandClaude Opus 5.5