The three post-1.2.5 CI commits that tie image publish to a published GitHub
Release and trim per-arch GHCR tags were missing from the changelog.
Co-authored-by: Cursor <cursoragent@cursor.com>
Run release.yml directly on release:published instead of a dispatcher
workflow. Remove per-arch tags from GHCR after the manifest merge so only
X.Y.Z remains visible to operators.
Co-authored-by: Cursor <cursoragent@cursor.com>
GitHub recorded release published at 16:44:21Z but no release.yml run
started. A thin release-on-publish workflow listens for publish and starts
release.yml via workflow_dispatch (always creates a run). release.yml now
checks out vX.Y.Z from the version input, not main HEAD.
Co-Authored-By: Cursor <cursoragent@cursor.com>
Bare git tag pushes no longer start the image build; publishing a GitHub
Release does. workflow_dispatch still requires an explicit X.Y.Z version.
Docs explain draft vs published releases and that deleting a tag on GitHub
converts a published release back to draft.
Co-Authored-By: Cursor <cursoragent@cursor.com>
Expand the doc-alignment entry to cover guide RBAC and restore Resync
wording; trim the P2 entry of claims superseded by that pass while keeping
its /license route-table addition with a cross-reference.
Co-authored-by: Cursor <cursoragent@cursor.com>
Fix architecture route RBAC, restore Resync, and session/password wording;
correct guide restore/session/PTR claims and README port-587 text; point
the CSRF ADR at authz.go for global route gating.
Co-authored-by: Cursor <cursoragent@cursor.com>
Move restore Resync under Fixed to match e70ba90, and extend the restore
test bullet with the resync boot step and TestResyncAfterRestoreHealsDriftedMaps.
Co-authored-by: Cursor <cursoragent@cursor.com>
Record the operator-guide restructure, merge the stale code-review plan
entry with its completion note, and align the restore Resync bullet with
the current backup package comment.
Co-authored-by: Cursor <cursoragent@cursor.com>
The section explained what a restore means conceptually but gave no
runnable commands. Add a step-by-step in-place restore, a
move-to-a-new-host variant, and the encrypted-backup decrypt-first
sequence, plus the actual version-mismatch error text so operators
know what to expect.
Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
Internal (non-operator) env vars were listed in guide.md's Environment
variables section even though the guide's own text says operators
aren't meant to touch them. They now live in architecture.md §
Configuration, the as-built/internals doc; guide.md keeps a one-line
pointer.
Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
Installation now reads Ports, Local trial, Initial setup, Full
deployment (with Fixed image tag nested under it), Environment
variables, Reverse proxy. The step-by-step production deploy and the
per-proxy TLS commands move from README's "Reference deploy" into the
operator guide; README keeps a short pointer instead of duplicating
the walkthrough.
Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
Splits mixed DNS-setup, Operations, Rate-limiting, and Backup sections
along the instance/domain boundary and fixes README anchors that
pointed at the old #dns-setup heading.
Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
Every phase (P0-P7) is closed, so the plan file goes per its own exit criteria; history lives in git and the CHANGELOG. The roadmap's recommended order returns to queue-retries and then inbound-relay, and the order notes in both plans stop pointing at the deleted file, as does the 1.2.5 changelog entry (same treatment as logrotate-mode.md).
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
The Fable review of the send-log authorization, tryAdmit, session create and app-delete ordering against security.md found no further issues: every send-log query path states its scope and an empty scope matches nothing, rejected rows cannot consume rate-limit budget, the milter's nil-registry fail-open is test-only, and each failure residue of the reordered delete fails safe. Outcome recorded in security.md's header and the plan; nothing added to the accepted risks.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
When CheckRestore accepts a backup manifest, the panel re-derives OpenDKIM tables and the Postfix sender map from SQLite on that first boot and reloads both daemons, so archive/database drift is healed before mail flows.
Co-authored-by: Cursor <cursoragent@cursor.com>
Close the remaining low-risk items from the full-tree review: rename the settings handler, query assigned domains in SQL, bound the login limiter map, collapse panel.js show/hide helpers, and soften DMARC copy that promised a future in-panel receiver.
Co-authored-by: Cursor <cursoragent@cursor.com>
IBM Plex WOFF2 files were shipped without the SIL OFL 1.1 text, NOTICE
told modifiers to edit layout.html for a Source URL that lives in
legal.go, and workflow_dispatch took GITHUB_REF_NAME as the version so a
run from main would publish ghcr.io/...:main.
Closes P5 of docs/plans/code-review.md. CHANGELOG updated.
Co-Authored-By: Cursor Grok 4.6 <noreply@anthropic.com>
Co-authored-by: Cursor <cursoragent@cursor.com>
Roadmap item and plan: show first retry delay, backoff cap and queue
lifetime on Mail queue and the delivery page, reading postconf -h once
at panel start so a manual override is visible. Explanation only; not
implemented yet. CHANGELOG updated.
Co-Authored-By: Cursor Grok 4.6 <noreply@anthropic.com>
Co-authored-by: Cursor <cursoragent@cursor.com>
Restore is not a code path in the panel — the operator extracts the archive
onto /data and starts the image — so it had no test. cmd/panel/restore_test.go
now performs that path in process: download a backup from a running panel
through POST /backup (plain and encrypted), unpack it the way tar -xzf does,
and boot a second panel on the result through run()'s own startup order
(CheckRestore, store.Open, newPanel, Start).
Covered: the restored panel shows the domain and journal the archive carried
and finds the DKIM key, sasldb2 and Postfix sender map where its configuration
says they are; the setup link is not reopened by a restore; a session that
predates the backup still works, as the guide documents; an encrypted download
restores identically; a data directory from another version is refused with
both versions named and the manifest kept for the retry.
serveHTTP is split so the composition it performs (newPanel) can be started
without binding a port. No behaviour change.
Closes the optional P4 item in docs/plans/code-review.md, and with it the
"HandleBackup POST untested" gap from the review's test section. CHANGELOG
updated.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Co-authored-by: Cursor <cursoragent@cursor.com>
- internal/web/auth/ratelimit_test.go: the login/setup limiter's ceiling,
per-address scope, window reset and the sweep that keeps finished
buckets out of memory.
- internal/web/auth/handlers_test.go: sign-in (session issued, refusals
that do not reveal which usernames exist, a lockout a correct password
cannot bypass, the setup hint before the first administrator) and the
one-time setup link (creates the first global administrator and then
closes, rejects a wrong or expired token, refuses credentials the panel
would not accept later, rate-limited).
- internal/web/handlers/authz_test.go: every global-only route answers a
domain administrator — and a request with no principal — with 404, with
a positive control so the table cannot pass on a handler that always
404s. This is the check that would have caught the send-log leak.
- test/e2e: CoreDNS pinned to 1.14.6 instead of latest; the level-1
failure message quoted RATE_LIMIT_MESSAGES_PER_IP=5 while the stand
sets 50.
docs/plans/code-review.md P4 checked off except the optional backup-boot
test; CHANGELOG updated. No production code changed.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Co-authored-by: Cursor <cursoragent@cursor.com>
- Deliveries fragment/pagination links urlquery-encode the domain/app
filters instead of splicing them into the query string raw.
- The four polled regions (deliveries rows, status, mail queue, system
log) carry aria-live="polite" so a screen reader announces refreshes.
- security.md accepted risks note that data-confirm prompts are
JavaScript-only and why that is acceptable.
docs/plans/code-review.md P3 fully checked off; CHANGELOG updated.
Co-authored-by: Cursor <cursoragent@cursor.com>
- .flash.error now renders on the danger surface instead of the success
one; RateLimitErr previously showed as green with red text.
- User delete goes through a confirmation page (GET/POST
/users/{uid}/delete), matching the domain-delete pattern, instead of a
plain submit button next to Save with no confirmation.
- Extracted the repeated DNS Host/Type/Value markup on a domain's page and
the duplicated Settings credentials form into shared partials. No
behaviour change.
docs/plans/code-review.md P3 checked off; CHANGELOG updated.
Co-authored-by: Cursor <cursoragent@cursor.com>
The CSRF ADR, guide.md, architecture.md route table, the internal/backup
package comment, roadmap/dmarc-reports admin.dmarc_report_email references,
the e2e coverage summary, and the setupManager/store package comments still
described the pre-1.2.0 single-administrator panel or stale implementation
details. Closes P2 of docs/plans/code-review.md; no behaviour change.
Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
Co-authored-by: Cursor <cursoragent@cursor.com>
The level-2 limiter counted stored plus in-flight messages and reserved its own slot in two critical sections, so SMTP sessions that overlapped could each take the last free slot; tryAdmit now does both under one lock. A session that cannot be written no longer yields a cookie the browser would carry while every request bounced to /login. Deleting an application clears its SASL account before its registry row, matching domain delete, so a saslpasswd2 failure leaves a retryable application rather than an account that still authenticates.
Co-authored-by: Claude Opus 5 <noreply@anthropic.com>
Co-authored-by: Cursor <cursoragent@cursor.com>
The release gate timed out waiting for status=sent because its HTML scrapers had drifted from the panel's badge markup; record the fix and the handler regression test.
Co-authored-by: Cursor <cursoragent@cursor.com>
The deliveries_rows template wraps status in badge spans, but the e2e gate still looked for bare td cells, so send_verify_dkim_and_status timed out even when mail was delivered and logged. Add a handler regression test so go test ./... catches this drift without Docker.
Co-authored-by: Cursor <cursoragent@cursor.com>
The Deliveries list narrowed the journal only when exactly one domain was assigned, so an administrator with none or with two or more read every domain's rows. The domain scope is now an IN constraint the store query carries, a filter that states no scope matches nothing, and the domain/app query parameters are validated against the principal before the query runs.
Co-authored-by: Claude Opus 5 <noreply@anthropic.com>
Co-authored-by: Cursor <cursoragent@cursor.com>
Capture the 2026-08-13 review as the active plan, queue it ahead of inbound-relay, and defer SQLite migration squash to 2.x with a user_version gate.
Co-authored-by: Cursor <cursoragent@cursor.com>
Drop the shared-baseline flex pin on .split cards; mismatched body
lengths made Export / Danger look worse than the plain stack.
Co-authored-by: Cursor <cursoragent@cursor.com>
Subgrid Domain settings rows, match Address mode and trusted-IP controls,
and pin Export/Danger buttons to a shared baseline in split cards.
Co-authored-by: Cursor <cursoragent@cursor.com>
Drop the app L1 backstop line, move trusted-IP help under the IP field,
shorten the domain rate-limit title, and align field-pair labels with DMARC.
Co-authored-by: Cursor <cursoragent@cursor.com>
Restore muted L1 backstop line, move domain limit help above Save, and drop
the duplicate Address mode field label.
Co-authored-by: Cursor <cursoragent@cursor.com>
1.2.4 stays as cut at the release commit; the L1 display and badge work
belongs in Unreleased until the next version cut (development.md).
Co-authored-by: Cursor <cursoragent@cursor.com>
Show the level-1 backstop as its own line and code-row, use st badges for
active/inactive domain and app limits, and fold the polish into 1.2.4.
Co-authored-by: Cursor <cursoragent@cursor.com>
Close Unreleased for the level-2 rate-limit semantics change and DNS Type
field height fix; pin compose and docs to 1.2.4.
Co-authored-by: Cursor <cursoragent@cursor.com>
Invert level-2 semantics so domain limits apply to every client IP and
application limits with trusted IPs raise the ceiling above the domain
(still capped by level 1). Panel shows L1, validates maxima, and documents
the model on Settings.
Co-authored-by: Cursor <cursoragent@cursor.com>
Close CHANGELOG [Unreleased] as 1.2.3; bump the compose pin and image references. Includes domain detail layout polish and Domains add-row.
Co-authored-by: Cursor <cursoragent@cursor.com>
Settings moves to /settings with a 308 redirect from /account; domains, Status, Users, and user forms get matching browser titles and h1 text; backup page title reflects domain import.
Co-Authored-By: Composer <noreply@cursor.com>
Co-authored-by: Cursor <cursoragent@cursor.com>
Close CHANGELOG [Unreleased] as 1.2.2; bump the compose pin and image references.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
Co-authored-by: Cursor <cursoragent@cursor.com>
Pair Machine|Processes, queue|cert, and sockets|hostname; fix .split
auto-margins so cards fill half the row; trim Status prose and machine
details; drop the page section index; note panel-docs on the roadmap.
Co-authored-by: Cursor <cursoragent@cursor.com>
Widen the status page and wrap the short queue/certificate cards in
.split so the stack is shorter; other checks stay full width.
Co-authored-by: Cursor <cursoragent@cursor.com>
Form-wide block labels and full-width inputs stacked the box above the
domain name; checkbox rows now share label.check and fieldset spacing.
Co-authored-by: Cursor <cursoragent@cursor.com>
Close CHANGELOG [Unreleased] as 1.2.1; bump the compose pin and image references.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
Co-authored-by: Cursor <cursoragent@cursor.com>
IBM Plex Mono sits low in its em square; top-heavy badge padding left
ok/warn sunk below the adjacent heading. Bottom padding is now heavier.
Co-authored-by: Cursor <cursoragent@cursor.com>
Global administrators manage every domain; the assignment checkboxes are
now hidden when that role is selected.
Co-authored-by: Cursor <cursoragent@cursor.com>
Two full silhouettes with staggered baselines read lopsided at 16 px; the
rear person is now a right-side crescent behind an aligned front silhouette.
Co-authored-by: Cursor <cursoragent@cursor.com>
The icon was a sun-with-rays (circle plus spokes), not a cog, despite the
1.2.0 note that Settings used a gear.
Co-Authored-By: Cursor <cursoragent@cursor.com>
The user form had Back to users at the bottom of the card; all drill-down
pages now use a shared back_link template with a structural test.
Co-Authored-By: Claude <Composer 2.5> <noreply@anthropic.com>
Co-authored-by: Cursor <cursoragent@cursor.com>
Global administrators get two cards in the .split layout; domain-scoped
users keep the single narrow form.
Co-Authored-By: Claude <Composer 2.5> <noreply@anthropic.com>
Co-authored-by: Cursor <cursoragent@cursor.com>
Close CHANGELOG [Unreleased] as 1.2.0; correct missing entries; bump the
compose pin and image references.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
Co-authored-by: Cursor <cursoragent@cursor.com>
Settings and the user form centred the title on the 48rem measure while
.card.narrow centred on 24rem, so their left edges drifted by 12rem. Cap every
direct child of main when a narrow card is present, without shrinking the
column, so navigation stays put and login/setup stay unchanged.
Co-Authored-By: Cursor <cursoragent@cursor.com>
Settings had been sharing the account silhouette with Users and the signed-in user label, so three different concepts read as the same icon. Settings now uses a gear, the user line the single-user icon, and Users a two-person group mark.
Co-Authored-By: Cursor <cursoragent@cursor.com>
Settings and the user form leave their heading at the left of the reading
measure while the card floats 12rem to the right of it. main > * caps every
child at 48rem and centres it; .card.narrow overrides the max-width down to
24rem but not the auto margins it never mentions, so the two centre on
different measures and half the difference is the offset.
Records what the question was actually about: not what is wrong with Settings
but what it shares with exactly one other page. Four templates use the narrow
card and eleven use a full-width one, which sorts every page into three groups
— full-width card, narrow card without navigation, narrow card with it — and
only the last has the problem. Settings and the user form are the only pages
that are both signed-in and built from a single narrow card; every other page
has one of those properties, never both.
Also records why the obvious fix is unavailable, so it is not tried a fourth
time: the shell centres the navigation and the page as a pair, so anything that
changes the column width is visible from the navigation (296px, measured). The
fix has to work inside the column.
Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
The send log was the one place in the panel where a status was bare text, next
to a delivery's own page where the same value is a badge, and a domain list
where the DNS verdict is one too — so a message's outcome looked like a
different kind of fact depending on which page you read it from.
The level travels on the row rather than being worked out in the template:
deliveryLevel is the one place that decides what a status means, the delivery
page already reads it, and a second mapping written into the template or the
stylesheet would have been free to drift from it. sent is ok, deferred a
warning, bounced and rejected errors, and queued unknown — nothing has gone
wrong, nothing has been reported yet.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Two fixes in one file, both from looking at the result.
Settings and the user form were narrowed to their card's width so their heading
would stop floating beside it. The shell centres the navigation column and the
page as a pair, so narrowing the page moved the navigation with it — 296px
sideways between Settings and Domains, which is the jump the width rules at the
top of this file exist to prevent, arrived at from the other direction. Every
way of aligning the heading with the card moves something else, so the geometry
goes back to what it was before the restyle, rendered against the previous
stylesheet page by page to confirm it. Those two headings are still not aligned
with their cards; that is worth fixing on its own terms rather than by making
the chrome move on every page.
And the badge text sat above the middle of its box: the badge inherited the
body's 1.5 line-height, and a lowercase word with no descender — ok, warn,
unknown — sits high in a line box that tall, which inside a border reads as
text stuck to the top. The box now hugs its line and the padding does the
centring, uneven on purpose to pay for the descender space the words rarely
use. The one that does, queued, keeps the same box.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
The compact button rule is written for controls that cluster — several beside a
value or inside a table cell — where 0.8rem/600 is what stops four of them
shouting at once. Sign out borrows that rule for its colours, and was taking
the type with it: in the navigation column it has no cluster to belong to, so
it stood under Settings among entries set 0.95rem/400 as the one smaller,
heavier word on the column, reading as a different kind of object rather than
as the last item of the list.
Type and padding go back to the entries'. The red, the border and the tinted
background stay — those are what say it acts, and the restyle only made the
mismatch easier to see.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
An image whose /etc/logrotate.d/mail is group-writable makes logrotate ignore
the file, so mail.log never rotates and grows until the volume does. The file
is 0644 in git and is widened on the way into the build context: COPY takes the
mode it finds, and an archive made from a checkout without POSIX permissions
carries the umask-widened one. Measured at 0644, 0664 and 0666 across three
images built on one host from the same commit range.
Written up rather than patched because three things are wrong and fixing the
visible one hides the rest: every COPY in the image trusts the build context,
logrotate exits 0 while ignoring the file so the rotate loop reports success,
and nothing anywhere notices that a rotation did not happen. Whether the
release workflow's own images are affected is stated as unverified — the
published image could not be pulled to check.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Deployed at 1.1.0-post.669f928: the policy is untouched and still admits the
fonts, each comes back as font/woff2 with a content ETag that answers a
matching If-None-Match with 304, and the signed-out page renders in Plex over
the network. The remaining open items all need a signed-in session with real
mail behind it.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Five things, each written down so the next pass does not rediscover them: the
send log's bare-text status and the mapping question that keeps it from being a
pure repaint, the 375px horizontal overflow that predates this work, the three
views only ever seen empty because they need a running Postfix, the CSP and
ETag checks that a file:// render cannot make, and font-display: swap, which
never had a cold load to show itself in.
Also records how the pages were reached at all — the stubs and seeded rows
without which half the panel does not exist on a development machine.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
The last cell in a row is the one the automatic table layout squeezes first,
and in the send log it holds a fixed two-word link: "Details" was being split
after "Detail" on every row. Holding that cell on one line costs the row
nothing and hands the width back to the status column, which was breaking
"deferred" into "deferre" and "d" for the same reason — the status gets the
same nowrap the timestamp already had, through a class of its own since the
cell had nothing to select on.
Both predate the restyle; they surfaced while screenshotting the send log
against a seeded journal.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
The stamp was approved on its own sheet and the panel kept the blue-on-cool-grey
it shipped with, so the mark read as pasted onto someone else's page. Brick
becomes the accent, the greys warm to paper, and the corners square up towards
the stamp's own edge.
Brick is two tokens rather than one. #7A3B2E under white text is 8.4:1, but as
text on the dark scheme's background it is about 2:1 — a link nobody can read.
So --accent-fill carries controls and --accent-text carries links and the active
navigation entry; in the light scheme they coincide, in the dark one they do not.
IBM Plex ships with the panel instead of being named and hoped for. The mark is
Plex converted to outlines, so a system stack left the panel's own name as the
only Plex on the page — and ui-monospace resolves to Consolas, SF Mono or DejaVu
Sans Mono depending on the machine, which laid the send log and mail.log tables
out differently for every operator. Three WOFF2 files, 76 KB, served from the
panel's own origin so default-src 'self' still covers them.
Two things changed shape rather than colour. Links had no rule at all and ran on
the browser's blue, invisible while the accent was blue and wrong the moment it
was not. And the column-narrowing that login and setup had by name now keys off
the narrow card itself, so Settings and the user form stop splitting their
heading from the card it belongs to.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
The mark was approved on its own sheet and the panel was left on the default
blue-on-cool-grey it shipped with, so the two read as different designs. Write
down what the panel takes from the proof — brick as the accent, warm paper,
IBM Plex self-hosted — and what it deliberately does not: the proof's mock puts
navigation in a dark top bar, which has nowhere to keep the per-page section
index the left column carries.
Records the decisions that cost something to reverse: three font files (76 KB)
served from the panel's own origin so the CSP stays default-src 'self', brick
split into a fill and a text value because #7A3B2E is unreadable as a link on
the dark scheme, and the test that outranks all of it — the accent must not be
mistakable for a status badge.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
The runtime stage copies LICENSE and NOTICE into /usr/share/doc/selfpost/
for AGPL-3.0 conveyance. Excluding LICENSE in .dockerignore broke clean
docker build once the cached layer was invalidated.
Co-authored-by: Cursor <cursoragent@cursor.com>
Close CHANGELOG [Unreleased] as 1.1.0; remove erroneous DMARC entries
retroactively copied into older sections; bump the compose pin and image
references.
Co-Authored-By: Cursor <cursoragent@cursor.com>
Lay out view, auth, validate, and handlers under internal/web while keeping
the cmd/panel API unchanged; update roadmap and changelog for web-split closure.
Co-authored-by: Cursor <cursoragent@cursor.com>
The panel now suggests policy-only DMARC by default, lets operators configure a default and per-domain report address, and DNS-checks hub _report._dmarc records. Future in-panel report ingestion is tracked as dmarc-reports in the roadmap.
Co-authored-by: Cursor <cursoragent@cursor.com>
Name the copyright holder, ship NOTICE with the image, serve the licence
from the panel footer on every page, and record the vendored htmx licence.
Co-authored-by: Cursor <cursoragent@cursor.com>
Close the [Unreleased] section as 1.1.0 and move the deploy pin, the README
trial command, and the guide's stated pin to that tag.
Corrections to the section while cutting it:
- The GHCR "unknown blob" retry was filed under Unreleased, but the commit
carrying it is what v1.0.0 points at. Moved into the 1.0.0 section.
- docs/plans/ and the roadmap's restructure into a 1.x+ tracker were missing
entirely; the plans were described only as "translated", which reads as if
they predated this release. Filed under Added.
- docs/product.md's reframing of the future line -- agreed 1.x+ extensions,
inbound relay targeting a MINOR by default, approval needed only for
candidates -- was unrecorded.
- The Account/Settings rename also touched the operator guide.
Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
The assets are baked into the binary, so their FS modification times are the
zero value and http.FileServer sent no Last-Modified. With no ETag either, a
static response carried no validator at all and the browser was free to guess
how long to keep it -- which is how the panel kept showing the previous favicon
in the tab well after the new mark was deployed.
Hash each asset once at startup and serve that as a strong ETag, with
Cache-Control: no-cache so the browser keeps its copy but revalidates. An
unchanged asset then costs a bodyless 304 and a changed one is picked up on the
next load.
Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
The question kept resurfacing while preparing the repository for publication,
so it is written down as settled rather than re-argued: SelfPost is written by
AI agents under a maintainer's direction and the project says so.
What that commits to is listed explicitly — the Co-Authored-By trailers, the
model routing table, the .cursor rules file, and the agent-facing process notes
in the roadmap all stay, and none of them is an oversight to be tidied away.
The reasoning is that with the trailers already in ~140 commits, removing the
surrounding material would not conceal authorship, only signal an attempt at
it.
The revisit condition is deliberately narrow: a conflict with the licence or a
downstream obligation, not a shift in convention.
Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
A reader looking for where to report a flaw had no path from the README: the
only security row led to docs/security.md, which is the design rationale, not
a reporting channel. That row is now "Security design", and a line under the
table sends vulnerability reports to SECURITY.md instead of the issue tracker.
development.md gains SECURITY.md in its user-facing deliverables table and
drops "internal" from the roadmap row, which stopped being true when the
roadmap was reframed for outside readers.
Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
The roadmap links straight into these files, so a reader following an item
landed in Russian one click after an English page. Translated in full; goals,
boundaries, done-when criteria and risks are unchanged in substance.
The model assigned to inbound-relay is dropped, as it was from the roadmap
item: model routing lives in development.md.
Figures and code references were checked against the tree while translating
rather than copied forward — internal/web is still 50 files, 25 .go, ~4300
lines, and requireAuth(authed) and sendLogData are still where domain-admin.md
says they are.
Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
roadmap.md is linked from the README table, so it is a page strangers land on.
Translated in full, with a note up front that none of it is a commitment:
there are no dates, the order is a recommendation, and an item can be dropped
once its plan is written. The note also points at product.md for the opposite
question — what the project deliberately will not do — so absence from the
roadmap is not read as a silent plan.
The model assigned to inbound-relay is dropped from the item; model routing
belongs in development.md, not in a statement of direction.
The README row described the file as internal and Russian. It is now neither.
Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
The 0.5.0 entry pointed at docs/security.md § "Резервная копия и экспорт
домена", a heading that no longer exists after the file was translated. The
name is updated to the current heading; the entry otherwise stands as written.
Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
security.md is linked from the README documentation table and now from
SECURITY.md, so a reader following either link landed in a Russian document
while everything around it was English. Translated in full; the requirements,
the accepted risks, and the CSRF ADR are unchanged in substance.
The reviewing model is no longer named in the text — that the pre-release
review ran, and when, is what a reader needs; who ran it is process detail
kept in development.md.
extract-cert.sh keeps its spec 10.3 quotation, translated. In sasl.go the
quotation from the closed plan is dropped rather than translated: rendered in
English it restated the sentence it hung off.
Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
A public repository with no stated disclosure channel routes a finder into
opening a public issue, which discloses a relay flaw to everyone the moment it
is filed. SECURITY.md points at the repository private vulnerability reporting
instead, with public@mixeme.ru as fallback, and states scope so operator-side
configuration (blocked port 25, missing PTR, proxy TLS) does not arrive as a
report.
No response time is promised: a deadline that cannot be honoured by a single
maintainer is worse than none. Silence is explicitly not a request for a
continued embargo.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
The nav entry, page heading, and browser title now read Settings. The
route, template name, and Active key stay `account`, so existing links
and bookmarks keep working.
Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
The import form used to ask users to tick "the file is encrypted" before
showing the password field, even though the server already decides purely
from the envelope magic bytes. Reveal the password field automatically for
a .spde file (hide it for .json) so the checkbox is no longer needed.
The navigation was a bar across the top that did not fit on one row — six page
entries and the session block against the panel's width — and had to be split
into two, costing the top of every page. It is now a column down the left edge:
one left edge to scan, the current entry marked down its leading edge, sticky so
it stays in view, and room under the entries for the current page's own
sections. Below the width the two columns need it lies back down into the same
wrapping rows as before; six entries need no drawer.
The section index is for the two pages long enough to need one — the domain page
(nine cards) and the status page (eight). Each card carries an id and the page's
template defines the list by overriding an empty "sections" block in the layout,
so a page that defines nothing renders no index. panel.js marks the section in
view, looking targets up by id on each pass so the status page swapping its
cards out every five seconds cannot leave it measuring boxes that have left the
document; the links themselves are plain fragment links and need no script.
Verified against the real pages rendered by a local panel at 1300px, 924px and
481px wide.
The status page answered "are the components running" but said nothing
about the machine underneath them, so a server slowed to a crawl by a
busy processor or one about to have Postfix OOM-killed looked entirely
healthy until the queue backed up.
internal/health/machine.go reads the kernel's counters in /proc: the
aggregate processor times and core count from /proc/stat, the load
average from /proc/loadavg, memory and swap from /proc/meminfo, and
per-interface byte counters from /proc/net/dev.
CPU busy time and network throughput are rates, so a MachineSampler holds
the previous reading and each call reports the difference — one shared
sampler on the Server, since a per-request one would never have anything
to subtract. A window longer than a minute only re-baselines: a page
opened after the panel sat idle would otherwise average that whole
stretch and present it as the current load.
Memory is derived from MemAvailable rather than MemFree, because Linux
spends every spare page on cache and MemFree would report a permanent
emergency. A fully busy processor (>=90%) warns and an exhausted machine
(>=97%) errors, both counting towards the page's headline verdict, since
either delays or kills the mail path. Throughput has no comparable
threshold — what counts as a lot depends on the link — so it is reported
and never graded. Loopback is excluded: that traffic is the container
talking to itself.
Like every other check here, an unreadable counter degrades to "unknown"
with an explanation instead of failing the page, so the panel still runs
outside Linux for development.
The usage bars are <meter> elements. The panel's CSP has no inline-style
exemption, so a bar's length has to travel on an attribute; the element
also grades its own colour from low/high/optimum, and the percentage is
printed beside it for anything that does not render meters.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Everything under Unreleased goes out as 0.5.0: the log-tailer bounce
mis-parse and offset-persistence fixes, the L2 rate-limit in-flight
accounting, the GitHub-only module path move, the code-review.md
retirement, the pre-release security pass (saslpasswd2 argv hardening),
and optional password encryption for backup/domain-export (Added, hence
a minor bump).
The heading is the version: there is no constant in the tree, only the
VERSION build arg that ldflags stamps into internal/buildinfo.Version,
so the image this deploys under is tagged from here.
Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
The review's plan is finished — phase 0 (bar the two release-commit steps),
1, 1.5, 2 and 3 are all closed — and what remained in the document was a second
copy of things that already live in architecture.md, security.md, roadmap.md or
the code comments: the GUI compromise table is in panel.css/panel.js/
middleware.go/handlers_auth.go, the single SQLite connection and the dual
cookie names are explained where they are implemented, the accepted gaps are in
security.md, and the model-routing table names progress.md and development.md
as its own source. A second copy of a fact is a place for it to go stale.
Four items were genuinely open and had no other home, so they moved to
roadmap.md rather than disappearing:
- splitting internal/web into subpackages (2.x) — with the reason to wait: the
flat package still reads at 47 files, and both 2.x features grow it, so the
cut is worth making before that growth, not now;
- a consolidated documentation index in the README (v1.x tail);
- the adaptive polling interval for a tab that is visible but idle — the hidden
case is already handled, and the remainder is explicitly allowed to end as
"decided not to";
- CONTRIBUTING.md, already moved to 2.x in the previous commit.
References retargeted: progress.md (7), roadmap.md (5), implementation-plan.md
(1). The CHANGELOG entries that cite the document are left as written — they
describe what happened at the time. The review text stays in git history at
aaf0711.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Codeberg is being retired as the project's public site, so every reference now
points at GitHub. That includes the Go module path (codeberg.org/mix/selfpost →
github.com/mixeme/selfpost): leaving an import path on a host that is going
away would break `go get` and `go install`, so this is not only a docs change.
Touches go.mod, test/e2e/go.mod, all imports, Makefile MODULE, the -ldflags
version stamp in build/Dockerfile and docs/development.md, the licence headers
in the SVG/HTML assets, and README (no more primary/mirror pair).
Comments no longer cite the archived specification. "spec 7.6.1", "spec 5.1"
and friends pointed into docs/archive/specification-v1.0.md, which is marked as
not a source of truth; each is now a reference to the live document that owns
the subject — architecture.md (with section), product.md, security.md or the
README. The review only asked for the 7.x refs (code-review.md § 4), but 4/5/6/
8/9 had the same defect, so they went too. Comments only, no behaviour change.
Also closes the remaining review items: architecture.md gained a Code layers
section with the layer diagram (A2), and TestParseDelivery gained the exotic
mail.log cases (§ 3).
Fixes a bug that last test found: the delivery-line pattern matched status=
greedily, taking the *last* occurrence on the line. Postfix appends the remote
server's reply verbatim, so a rejection whose reply quoted "status=sent" was
filed as a delivered message in the send log. It now takes the first status=
after the recipient, which is the real field.
R7 (CONTRIBUTING.md) moved to roadmap 2.x — one developer, no external PR flow,
so the file would have no audience yet. R1 (compose image tag) and the git tag
stay in roadmap § v1.x as the release-commit steps.
gofmt/go vet clean on both modules; go test ./... green except the three known
Windows-only failures (file perms, backslash paths, renaming an open file).
Not exercised on the dev server — no Docker locally.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
- logtail: persist the read position (offset + fingerprint of the log's
first 512 bytes) in a new logtail_state table (migration 0003) and
resume from it on start, so delivery lines written while the panel was
down are parsed instead of skipped and their send-log rows no longer
stay "queued" forever. Fingerprint mismatch (rotated/recreated while
down) reads the file from the start — re-parsing is idempotent; a
first-ever start with nothing stored still begins at end-of-file.
Writes are throttled to one per 5s, forced on rotation and shutdown.
- milter: count messages that passed the level-2 check but have not
reached the send log yet (internal/milter/inflight.go), so concurrent
SMTP sessions cannot each spend the same last slot. A literal
count+insert transaction, as the review suggested, is not possible:
the count happens at MAIL FROM and the insert at end-of-message.
Reservations are released after the insert, on ABORT, and after a
10-minute TTL — a client that drops mid-transaction must not be able
to hold a slot, since the limiter is fail-open by design.
Docs: architecture.md (log tailer, persistence, L2 counting),
security.md and roadmap.md (restart gap closed, container recreate
remains), CHANGELOG, progress.md, code-review.md.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
- panel.js: skip HTMX polling requests while the tab is hidden, via
htmx:beforeRequest rather than htmx's eval-based trigger filter (the
panel's CSP allows no unsafe-eval).
- panel.css: replace all dark-mode !important overrides with CSS custom
properties reassigned once under prefers-color-scheme: dark.
- panel.css: consolidate the duplicate main{max-width} rule.
Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
Both secret-bearing downloads can now be sealed with a password. Unticked, the
forms produce exactly the files they did before.
- internal/secretfile: envelope format — magic/type/scrypt params/salt/nonce
prefix header, then 64 KiB AES-256-GCM chunks each authenticated with the
header, its counter and an end-of-stream flag, so truncation, reordering and
tampering fail to open instead of restoring a plausible prefix. Streams both
ways, so a full backup never sits in memory.
- Panel: "Encrypt with a password" checkbox on the full-backup and
domain-export forms (shared partial, toggled from panel.js — no inline
script); domain import detects an encrypted export by magic bytes, not by
extension, and asks for the password.
- selfpost-backup: writes .spbk when given a password and converts one back
with -decrypt, which a restore needs. The password comes from
SELFPOST_BACKUP_PASSWORD or -password-file, never argv.
- Docs: README, security.md (+ accepted risk: encryption stays opt-in),
architecture.md, progress.md, CHANGELOG.
Verified locally: panel-encrypted archive decrypts through the CLI and unpacks;
wrong password and password mismatch are refused; UI checked in a browser.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
mixandCursor