Testing story: app-level broadcast audits (global ActionCable.server) can be silently buffered by ActionCable::TestHelper, and 4.1.x gives tests no synchronous flush
Nessuno ha ancora preso questa issue.
Valutazione
- Difficoltà
- 3/5
- Tempo stimato
- 1-2 giorni
- Idoneità per principianti
- 30/100
- Tipo di issue
- Documentazione
- Chiarezza
- Abbastanza chiara
- Stato di attività
- Attiva
- Ambito
- documentation, testing
Direzione di ricerca
Start with the README testing section and the inline_broadcasts mode in solid_cable_test.rb, then trace how ActionCable::TestHelper swaps the pubsub adapter to SubscriptionAdapter::Test on the global ActionCable.server. The documentation note should explain the swap and how to reset it or use a dedicated server. The optional flush_broadcasts helper for BatchedBroadcaster needs maintainer agreement on direction before any code is written.
Scritto dal modello di indicizzazione a partire dal testo della issue.
Descrizione
(Cross-reference: we filed hotwired/turbo-rails#793 about the
auto-inclusion mechanism; this one is the Solid Cable side of the same
finding: what it means for Solid Cable users and what could be added.)
Context
A Rails 8 app auditing Solid Cable's core promise in a test ("a broadcast
is a DB INSERT into solid_cable_messages") wrote this test:
test "broadcast lands in solid_cable_messages as a DB row" do
ActionCable.server.broadcast("solid_cable_audit", payload)
row = poll_until { SolidCable::Message.order(:created_at).last } # up to 30s
assert_not_nil row
end
and watched it fail intermittently, with the row never appearing at all,
while the same test passed standalone and in other suite orderings.
Root cause (full instrumented timeline in hotwired/turbo-rails#793):
turbo-rails auto-includes Turbo::Broadcastable::TestHelper (which pulls
in ActionCable::TestHelper) into every ActiveSupport::TestCase
mid-suite. ActionCable::TestHelper#before_setup then swaps the global
ActionCable.server's pubsub to the buffered SubscriptionAdapter::Test
adapter for every test. Solid Cable is never consulted, no row is ever
written, and nothing logs or raises.
Why Solid Cable's own suite doesn't see this
Your adapter tests construct their own
ActionCable::Server::Base.new per test with their own cable config, so
the global-server hijack can't touch them, but it bites every application
test that uses ActionCable.server directly, which is the natural way an
app verifies "broadcast → DB row". Since Solid Cable is the Rails default
pubsub, this interaction is likely to keep surprising users.
The async-writer side
Even with the real adapter active, BatchedBroadcaster (4.1.x) gives tests
no synchronous flush: we observed the writer thread's first flush lagging
past 10s once on a cold process (both runtimes), which forced a 30s poll
budget. Your own suite engineers around exactly this with a poll helper
(wait_for_messages, sleep 0.001 until ... under a 5s Timeout).
Main's inline_broadcasts mode (writes synchronously inside the Rails
executor, verified against your solid_cable_test.rb) is a big
improvement, but note it does not fix the buffering problem above:
ActionCable::SubscriptionAdapter::Test never reaches SolidCable at all,
so no Solid Cable-side mode can help while the swap is active.
What we'd like
- Documentation (README testing section): when testing broadcasts
against the DB through the globalActionCable.server, beware
ActionCable::TestHelperswapping the pubsub (auto-included by
turbo-rails); reset the adapter (ActionCable.server .instance_variable_set(:@pubsub, nil)) or construct a dedicated server.
This note is the part that helps users today:inline_broadcasts
(see above) does not cover the swap, because a broadcast that is
buffered by the Test adapter never reaches SolidCable. - Optionally a small
SolidCable::Testinghelper (e.g. a
flush_broadcaststhat drains the currentBatchedBroadcasterqueue,
or guidance to enableinline_broadcastsin the test environment)
so app-level audit tests don't need hand-rolled 30s polls.
Both are small; happy to send PRs for either once the direction is OK'd.
- Lingua principale
- Ruby
- Stelle
- 348
- Fork
- 33
- Metriche di merge delle PR
- Nessuna PR unita negli ultimi 30g
Preparare l'ambiente
Questo progetto non fornisce container di sviluppo, Dockerfile né guida per i contributori, quindi l'ambiente è a tuo carico: parti dal suo README e consulta la nostra guida al primo contributo per i passaggi generali.
Come iniziare
- Leggi tutta la issue e poi la guida ai contributi del progetto.
- Commenta sulla issue per dire che te ne occupi tu — evita che due persone facciano lo stesso lavoro.
- Fai un fork del repository e lavora su un branch.
- Apri una pull request che faccia riferimento al numero della issue.
Altre issue di rails/solid_cable
-
Difficoltà 5/5 Più di una settimana Idoneità per principianti 30/100
rails/solid_cable#97 · 1 commento ·
Tutte le issue di rails/solid_cable
Issue simili
-
Bug Ruby Waiting for: Product Owner
Difficoltà 2/5 1-3 ore Idoneità per principianti 78/100
getsentry/sentry-ruby#3228 · 1 commento ·
I maintainer di solito rispondono entro 1 giorno
-
Difficoltà 2/5 1-3 ore Idoneità per principianti 82/100
I maintainer di solito rispondono entro 1 giorno
-
開発用DockerイメージにPlaywrightのシステム依存を追加するForse già presa @momocus l’ha presa oggi. Aperta
Difficoltà 1/5 Meno di un'ora Idoneità per principianti 85/100
sakazuki-org/sakazuki#1453 ·
I maintainer di solito rispondono entro 3 giorni
-
Difficoltà 2/5 1-3 ore Idoneità per principianti 75/100
-
Difficoltà 2/5 1-3 ore Idoneità per principianti 76/100
jenkins-infra/jenkins.io#9615 · 1 commento ·
I maintainer di solito rispondono entro 1 giorno