Re-record the integration test's Jellyfin and TMDB traffic (1.5.3)

joeyr commented

2026-10-05 22:48:02 -05:00

Owner

The integration test runs again. Its recorded Jellyfin traffic predated the fork's own requests (1.0.3's type filter, 1.4.1's library lookup). Replay compares whole URLs, so it missed, and the run retried until the test timed out.

testdata/fixtures/jellyfin.yaml, tmdb.yaml: re-recorded against upstream's test Jellyfin (testdata/jellyfin, now 12.2.0), set up through its startup API. The recording run shared that container's network, so the recorded host stays localhost:8096. Neither key is in the cassettes (the hook drops Authorization).
newsletter_integration_test.go: FakeClock moves to 2026-10-06 12:00 UTC (the build day), and the subject month and "Added on" assertions follow.
The suite passes twice with -tags integration from the cassettes alone, with the test Jellyfin removed. Delivery went to the test's own Mailpit, so no real email was sent.
CLAUDE.md: the verified run command and re-record steps, and two corrected claims (unit tests don't use go-vcr; testcontainers starts Mailpit, not Jellyfin).
VERSION 1.5.3, CHANGELOG.

🤖 Generated with Claude Code

The integration test runs again. Its recorded Jellyfin traffic predated the fork's own requests (1.0.3's type filter, 1.4.1's library lookup). Replay compares whole URLs, so it missed, and the run retried until the test timed out. - `testdata/fixtures/jellyfin.yaml`, `tmdb.yaml`: re-recorded against upstream's test Jellyfin (`testdata/jellyfin`, now 12.2.0), set up through its startup API. The recording run shared that container's network, so the recorded host stays `localhost:8096`. Neither key is in the cassettes (the hook drops `Authorization`). - `newsletter_integration_test.go`: `FakeClock` moves to 2026-10-06 12:00 UTC (the build day), and the subject month and "Added on" assertions follow. - The suite passes twice with `-tags integration` from the cassettes alone, with the test Jellyfin removed. Delivery went to the test's own Mailpit, so no real email was sent. - `CLAUDE.md`: the verified run command and re-record steps, and two corrected claims (unit tests don't use go-vcr; testcontainers starts Mailpit, not Jellyfin). - VERSION 1.5.3, CHANGELOG. 🤖 Generated with [Claude Code](https://claude.com/claude-code)

joeyr added 1 commit

2026-10-05 22:48:02 -05:00

Re-record the integration test's Jellyfin and TMDB traffic (1.5.3)

CI / gate (pull_request) Successful in 1m3s

Details

CI / lint (pull_request) Successful in 57s

Details

CI / image (pull_request) Has been skipped

Details

d876302144

TestJellyfinNewsletter replays go-vcr cassettes, matched on the whole
URL. Those predated the fork's requests (1.0.3's IncludeItemTypes,
1.4.1's Path field and /Library/VirtualFolders), so replay missed, the
recorder tried a real Jellyfin at localhost:8096, and 1.4.0's retry
waited until the test timed out.

The cassettes are re-recorded against upstream's test Jellyfin
(testdata/jellyfin, now 12.2.0), set up through its startup API rather
than the browser, with the recording run sharing that container's
network so the recorded host stays localhost:8096. FakeClock moves to
2026-10-06 12:00 UTC, the build day, so the same items are new, and the
subject month and "Added on" assertions follow. Neither key is in the
cassettes. The suite passes twice with the integration tag from the
cassettes alone.

CLAUDE.md now has the verified command and the re-record steps, and
corrects two claims: unit tests don't use go-vcr, and testcontainers
starts Mailpit, not Jellyfin.

Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>

joeyr merged commit 32058afed6 into main

2026-10-05 22:50:30 -05:00

joeyr deleted branch test/rerecord-integration-fixtures

2026-10-05 22:50:30 -05:00

joeyr referenced this pull request from a commit

2026-10-05 22:50:32 -05:00

Release 1.5.3: re-record the integration test's Jellyfin and TMDB traffic (#14)

Rows
Columns

Re-record the integration test's Jellyfin and TMDB traffic (1.5.3) #14