also looking at this
fix(ci): give the repair loop the guard it already had
#1407
10 files
+229
-34
| @@ -80,21 +80,27 @@ features: | ||
| 80 | 80 | details: "Every run posts an outcome: what it classified, what it did, and — when it declined — why the fix was not the bot's to make." |
| 81 | 81 | --- |
| 82 | 82 | |
| 83 | ## From the thread | |
| 83 | ## From a failing run | |
| 84 | 84 | |
| 85 | ```text | |
| 86 | @buddy fix-ci | |
| 87 | ``` | |
| 85 | Repair is driven by the run that failed, because the diagnosis is a function of | |
| 86 | that job's log. The workflow generated by `buddy setup` wires this up: when a | |
| 87 | run on a Buddy branch fails, the `fix-ci` job reads its log and reports back on | |
| 88 | the pull request. | |
| 88 | 89 | |
| 89 | The most common way to reach it. The command is permission-checked, so a commenter without write access cannot make Buddy push to your branch. | |
| 90 | A comment cannot say *which* run failed, so `@buddy fix-ci` in a thread replies | |
| 91 | pointing at that job rather than guessing. | |
| 90 | 92 | |
| 91 | 93 | ## From the CLI |
| 92 | 94 | |
| 93 | 95 | ```bash |
| 94 | buddy fix-ci --pr 128 | |
| 95 | buddy fix-ci --pr 128 --dry-run # classify and report, change nothing | |
| 96 | buddy fix-ci --run-id 12345 --pr 128 | |
| 97 | buddy fix-ci --run-id 12345 --pr 128 --dry-run # classify and report, change nothing | |
| 96 | 98 | ``` |
| 97 | 99 | |
| 100 | `--run-id` supplies the log to diagnose. `--pr` is where the report goes, and | |
| 101 | also where the attempt counter lives — without it Buddy cannot tell a first | |
| 102 | attempt from a fourth. | |
| 103 | ||
| 98 | 104 | ## From a workflow |
| 99 | 105 | |
| 100 | 106 | ```yaml |