pixa's CLAUDE.md says an existing test changes only with sneak's approval. On 15 October he approved two such changes, each a test that pinned the behaviour a filed bug calls wrong (#254 (comment), #242 (comment)). Three more bug fixes now wait on the same kind of approval:
#253 (comment) (a failed first-bytes check is kept in the negative cache, so one test case expecting a second fetch changes);
Question for sneak: may a fix change an existing test without asking you each time, when that test pins exactly the behaviour the filed bug says is wrong, and the change moves it to the fixed behaviour and nothing else? Any other change to an existing test would still come to you.
Recommendation: yes. Each such change is the bug fix itself, its reviewer checks that the test was changed only that far, and the three fixes above would land without waiting. If yes, CLAUDE.md gets that one sentence.
Model: opus-5-5
pixa's `CLAUDE.md` says an existing test changes only with sneak's approval. On 15 October he approved two such changes, each a test that pinned the behaviour a filed bug calls wrong (https://git.eeqj.de/sneak/pixa/issues/254#issuecomment-142374, https://git.eeqj.de/sneak/pixa/issues/242#issuecomment-142377). Three more bug fixes now wait on the same kind of approval:
- https://git.eeqj.de/sneak/pixa/issues/253#issuecomment-142424 (a failed first-bytes check is kept in the negative cache, so one test case expecting a second fetch changes);
- https://git.eeqj.de/sneak/pixa/pulls/278#issuecomment-143931;
- https://git.eeqj.de/sneak/pixa/pulls/279#issuecomment-143984.
**Question for sneak:** may a fix change an existing test without asking you each time, when that test pins exactly the behaviour the filed bug says is wrong, and the change moves it to the fixed behaviour and nothing else? Any other change to an existing test would still come to you.
**Recommendation:** yes. Each such change is the bug fix itself, its reviewer checks that the test was changed only that far, and the three fixes above would land without waiting. If yes, `CLAUDE.md` gets that one sentence.
Model: opus-5-5
sneak
was assigned by clawbot2026-10-15 22:53:52 +02:00
Plan: one sentence in AGENTS.md (where the agent rules moved from CLAUDE.md), in the rule on changing existing tests: a bug fix may change an existing test without asking when that test pins exactly the behaviour the filed bug calls wrong and the change moves it to the fixed behaviour and nothing else; any other change to an existing test still needs sneak's approval. Its TODO.md entry goes in the same commit. One PR to next, gated by an independent reviewer.
Model: opus-5-5
Plan: one sentence in `AGENTS.md` (where the agent rules moved from `CLAUDE.md`), in the rule on changing existing tests: a bug fix may change an existing test without asking when that test pins exactly the behaviour the filed bug calls wrong and the change moves it to the fixed behaviour and nothing else; any other change to an existing test still needs sneak's approval. Its `TODO.md` entry goes in the same commit. One PR to `next`, gated by an independent reviewer.
Model: opus-5-5
Blocking a user prevents them from interacting with repositories, such as opening or commenting on pull requests or issues. Learn more about blocking a user.
pixa's
CLAUDE.mdsays an existing test changes only with sneak's approval. On 15 October he approved two such changes, each a test that pinned the behaviour a filed bug calls wrong (#254 (comment), #242 (comment)). Three more bug fixes now wait on the same kind of approval:Question for sneak: may a fix change an existing test without asking you each time, when that test pins exactly the behaviour the filed bug says is wrong, and the change moves it to the fixed behaviour and nothing else? Any other change to an existing test would still come to you.
Recommendation: yes. Each such change is the bug fix itself, its reviewer checks that the test was changed only that far, and the three fixes above would land without waiting. If yes,
CLAUDE.mdgets that one sentence.Model: opus-5-5
clawbot referenced this issue2026-10-19 06:12:53 +02:00
yes
Plan: one sentence in
AGENTS.md(where the agent rules moved fromCLAUDE.md), in the rule on changing existing tests: a bug fix may change an existing test without asking when that test pins exactly the behaviour the filed bug calls wrong and the change moves it to the fixed behaviour and nothing else; any other change to an existing test still needs sneak's approval. ItsTODO.mdentry goes in the same commit. One PR tonext, gated by an independent reviewer.Model: opus-5-5
The sentence is in #306, waiting for review.
Model: opus-5-5