[dhongu]

This PR has two commits:

• 5a036336 [MIG] queue_job_cron_jobrunner: Migration to 19.0 — the same change as [19.0][MIG] queue_job_cron_jobrunner: migrate #876 (manifest + not-installable exclude list + regenerated README), taken from my earlier [19.0][MIG] queue_job_cron_jobrunner: Migration to 19.0 #979, which was byte-identical to [19.0][MIG] queue_job_cron_jobrunner: migrate #876. [19.0][MIG] queue_job_cron_jobrunner: migrate #876 cannot pass the CLA check as its commit author has not signed the CLA, so this PR carries the migration with a CLA-signed commit, rebased on current 19.0;
• 3572e102 [FIX] queue_job_cron_jobrunner: honour RetryableJobError seconds — the actual fix described below.

If #876 gets merged first, I will rebase and drop the migration commit.

Problem

When a job raises RetryableJobError, the cron runner of queue_job_cron_jobrunner always postpones it by a hardcoded 5 seconds. It ignores both RetryableJobError(seconds=N) and the retry_pattern of the job function, while the standard job runner (queue_job controller, _try_perform_job) honours them. A job that asks to be retried in, say, 10 minutes (rate-limited API, lock held elsewhere) is retried every 5 seconds instead, burning its max_retries in under a minute.

Fix

Mirror the controller's behaviour:

• RetryableJobError(seconds=N) → postpone by N seconds;
• no seconds → seconds=None, so Job.postpone applies the job function's retry pattern;
• neither set → keep the current 5 seconds delay (no behaviour change for existing setups).

Tests

Three new tests in tests/test_queue_job.py (explicit seconds, retry pattern, default). Without the fix, the first two fail; with it, all module tests pass on 19.0.

A towncrier fragment is included (readme/newsfragments/); no version bump, left to the bot.

Assisted-by: Claude Opus 5.5

[oca-cla-bot]

All commit authors signed the Contributor License Agreement.

[dhongu]

@ivantodorovich (addon maintainer), @espo-tony, @yelizariev — since you approved / tested #876: this PR carries the same queue_job_cron_jobrunner 19.0 migration (with a CLA-signed commit, rebased on current 19.0) plus a small fix so the cron runner honours RetryableJobError(seconds=...) and the job function's retry pattern, with tests. CI is green.

Could you have a look and, if it's fine for you, approve? A merge by a maintainer would then unblock the addon on 19.0. Thanks!

Assisted-by: Claude Opus 5.5