Fixed in 9d0a32c: REVIEWED_EXTENSIONS and reviewer_bad_extensions now precede pr_agent_settings.
Answer to the review of a59d494 (YELLOW):
- YELLOW
werkrevisor/review.py:71(settings read before they are applied): fixed in fdc6231, thread resolved. - YELLOW spec Scenario#35.01 `implementa…
Fixed in fdc6231: the So-that names what the model reviews (dependencies, repositories, plugins); title and Then no longer speak of the prompt.
Fixed in fdc6231: reviewer_bad_extensions reads the list once from language_extensions.toml shipped with PR-Agent, no longer from the live settings; `test_reviewability_does_not_follow_changed_…
Fixed in 537b48a: the scenario about the status context is gone, since the machine config on mih09 is outside this repository; the Solution names that change. The build of this PR reported…
Answer to the review of 8c244e0 (YELLOW):
- YELLOW risky paths (
.werkrevisor.toml, the Gradle files): renamed only, plus the version bump; needs the human review. - YELLOW spec Scenario#34.01…
Fixed in 537b48a: SettingsIntegrationTest sets the WERKREVISOR_STATISTICS_* names and checks that port, address and tokens arrive; it failed with a typo in the placeholder of the read tokens.…
Answer to the review of aa33df4 (YELLOW):
- YELLOW spec Scenario#33.05 and hint "contradicts the test and the code": fixed in d9b659a; the scenario now has a table for the resolved and the open…
Answer to the review of c3ad49e (ORANGE):
- ORANGE
werkrevisor/cli.py:499-500(HTTP 500 as refusal) andwerkrevisor/cli.py:455-456(edit before publishing): repeated findings, not changed,…
Fixed in aa33df4: test_a_green_review_refused_with_http_500_approves_without_line_comments.
Answer to the review of 0fcbb0f (ORANGE):
- ORANGE
werkrevisor/cli.py:499-504(reused comment not in sync after a 500): its text is lost, since 0.20.1 reused my reply 1458. From the title: an…
Not changed: the notes are the constants MERGE_SCHEDULED and MERGE_ALREADY_SCHEDULED, which _schedule_merge returns as they are; a change of the wording changes the constant and so the…
Not changed. _open_bot_line_comments reads only the comments of reviews submitted by the bot, so a bot reply in a thread started by a human, which Gitea lists under the human's review, is never…
Answer to the review of eeb7934 (ORANGE):
- ORANGE
werkrevisor/cli.py:430-432(a bot reply would be reused and edited): not changed, reasons in the thread; only bot reviews are read, and the…
Answer to the review of a356441 (YELLOW):
- YELLOW spec Scenario#33.07 and hint "contradicts the code on the cause": reworded in eeb7934; the scenario starts from the finding beyond the file,…
Not changed: the same point as the threads at cli.py:715 and cli.py:453; the result comment must know whether the edit failed, so the edit comes first. Left open for the human reviewer.
Answer to the review of 3d0fbbc (YELLOW):
- YELLOW
werkrevisor/cli.py:464-477(edit of line comments on real Gitea): verified with an unchanged edit of line comment 1406, HTTP 200; resolved. -…
Not changed, intended: a transient 500 still gets a second POST without line comments; if that fails too, the run ends with ERROR. Otherwise the findings are in the result comment, so nothing is…