queue-table: Outbound › Queue as the drawn table, Inbound › Queue built

Owner, 2026-10-18: "Фиаско, что очередь не была сделана. Как мы это
допустили? Исправляем!" The mockups out-queue and in-queue (design 629de5e,
guards 8525e8e).

- postfix: postqueue -j gives each message's size and each recipient's
  delay_reason (src/postqueue/showq_json.c, format_json(): "delay_reason"
  only when LEN(why) > 0); QueueListing reads it with the page's 5 s,
  QueueMessages keeps the notifier's 30 s.
- view: one Queue page type for both directions (queue.go), one polled box
  (queue_body.html): a row per recipient, oldest first, cut at 200 with a
  foot; the head "N messages · size", "Refreshes on its own" while empty;
  Last reply is the delay reason, or where the message is (active, hold,
  incoming) before there was one. Inbound has no Details: an inbound
  message has no page of its own.
- handlers: one read of the queue, split by the recipient's domain against
  the inbound domains; an outbound recipient is linked to its send-log row
  (store.SendLogRefsByQueueID, through idx_send_log_queue_id, plan pinned in
  TestSendLogQueriesUseTheirIndexes), matched without regard to case, else
  the message's first row. Both pages are the global role's; the inbound
  one 404s with the relay off. Routes /inbound/queue (+ /fragment), menu,
  touch menu and sibling strip.
- Overview: Queues · Both empty · outbound and inbound with the relay on.
- Help: the queue topic rewritten, an Inbound queue topic (word for word with
  the mockup; domain administrators do not see it, as with the lists).
- TestUnbuiltScreensAreOnTheRoadmap and its rule in development.md: an
  exempted screen must be named in roadmap.md; queue-table leaves the
  roadmap; guide, architecture, panel-redesign.md updated.

Tests: go vet; go test ./internal/web/... ./internal/store ./internal/postfix
./internal/notify pass.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
This commit is contained in:
mix
2026-10-18 23:17:14 +03:00
co-authored by Claude Opus 5.5
parent 8525e8ee8d
commit 6db3704711
34 changed files with 893 additions and 246 deletions
-20
View File
@@ -30,7 +30,6 @@ directory).
|---|---|---|---|---|
| security-audit-2026-10-17 | Stage H of the 2026-10-17 security audit: the test server's host (S1, S2, S5) and its DNS (S3) | stages 0-6 shipped `2.14.4`, the product side of S3 `2.15.0`; stage H the owner's | 27/29 | [security-audit-2026-10-17.md](security-audit-2026-10-17.md) § Stage H |
| go-milter-upstream | An issue and a PR for go-milter: its header map outlives a message and is unbounded | prepared 2026-10-16, the owner posts | — | [third_party/go-milter/PATCHES.md](../third_party/go-milter/PATCHES.md) |
| queue-table | Outbound › Queue as the table the mockup draws, from `postqueue -j`; an Inbound › Queue (`in-queue`) | candidate, undecided since 2026-10-09 | — | — |
After a context reset, pick an item marked `agreed` or `in progress`, then work
the **Implementation checklist** in its linked plan. The `Progress` column above
@@ -93,25 +92,6 @@ option), `third_party/go-milter` goes and `go.mod`'s `replace` with it.
---
## queue-table
**Problem:** the 2.0 mockups draw Outbound › Queue as a table and an
Inbound › Queue (`in-queue`); the redesign kept both out as decisions of
their own (owner, 2026-10-09). Outbound › Queue shows `postqueue -p`'s own
listing in a log pane, and the mockup's `in-queue` is drawn but not built
(`deferredScreens`, `outlineExempt` in the guard tests).
**Goal:** the outbound queue as a table from `postqueue -j` - a row per
message with its sender, recipients, age and last reason - and, if the owner
wants it, the same for mail port 25 holds for an upstream.
**Recommendation:** the outbound table first, read-only - about 4 h with its
tests, since the notifier already reads `postqueue -j` for each message's
arrival time; the inbound queue and any action on a row (retry, delete) only
if the owner asks for them, each drawn mockup-first.
---
## Dropped, with the condition for return
Items the owner took off the roadmap. Each names what would bring it back;