Rolling back ran the container from the local tag, so the demo recorded
`bulwark-demo:previous` as its image — and that tag is moved by the very
next deploy. Run the id instead; the tag stays only to keep the image from
being pruned.
The demo container followed ghcr.io/bulwarkmail/webmail:latest through
Watchtower, so a release reached it silently and there was no way to see,
trigger or undo it. It is now a deployment of this repo:
- .github/workflows/deploy-demo.yml runs after "Publish Docker Image on
Release" succeeds, or on demand with a tag
(`gh workflow run deploy-demo.yml -f tag=1.10.0`, or action=rollback).
It only talks to the server; nothing is built.
- deploy/demo-deploy.sh is the forced command behind the deploy key
(/usr/local/bin/bulwark-demo-deploy): pulls the tag, recreates the
container with the demo mounts, checks /api/health and goes back to the
image it replaced if that fails. A pre-release moves no :latest, so the
deploy after one finds the same image and does nothing.
Watchtower has been recreated without bulwark-demo; it still updates the
two rbm.systems containers.
Keep the latest Docker tags on the stable release when a GitHub release is marked as a pre-release, so auto-updating deployments on :latest do not pick up beta builds.
JMAP thread ids are only unique within their account - Stalwart hands out
per-account counters such as "b" - so the cross-account views merged
unrelated conversations that happened to share an id, and every
client-side thread map (expansion, cache, loading flag, counts,
virtualizer key) mixed them up the same way.
Give ThreadGroup an account-scoped `threadKey` derived from the email's
source stamps and key grouping and all thread state by it. `threadId`
stays the raw id for Thread/get, which #281 already routes to the owning
account. fetchThreadEmailCounts now asks each source account for its own
threads instead of one client for every id. Single-account views carry no
stamps, so their key is the bare id and nothing changes there.
Resolves the two conflicts in components/calendar/event-modal.tsx created
by #967 (RSVP controls), both keep-both:
- #967's canRsvp memo and #986's resolveContactName hook are both kept;
they are independent and were simply inserted at the same point.
- #967's RsvpBar component is kept, and StatusBadge takes #986's
signature ({ status, t }) since #986 drops the isOrganizer prop and
updates both call sites to match.
Non-admins usually may not read their own Account object (no
sysAccountGet), and fetchPrincipal was written to treat that as the
normal case. It only matched the word "forbidden" in the message, though,
while Stalwart answers with a `forbidden` method error whose description
reads "You are not authorized to perform this action" - so every calendar
load still logged an error and left an error message in the store.
Classify by the method error type (and HTTP 403), keep the wording checks
as fallback, and log the refusal at debug level only.
Since the freely scrolling views (#759) the day, week and month views are
positioned by their own focus rather than by the selected date, and a grid
click deliberately only marks a day that is already on screen. The mini
calendar shared that handler, so picking a day there highlighted it but
left the view where it was. Give it its own handler that jumps to the day
(keeping the mobile month-to-day switch), the same path the prev/next/today
controls and deep links use.