mirror of
https://github.com/kuhyx/testsAndMisc.git
synced 2026-11-19 17:44:23 +01:00
Move the trace-harness detail into the verification doc
Adding it to the handoff pushed that file to 280 lines -- the handoff is not exempt from the cap it exists to enforce. The usage, the two caveats (stubs always exit 0; unstubbed binaries run for real) and the proof it catches both bug classes now live in docs/shell-split-verification.md, which is the file dedicated to exactly this. The handoff keeps a pointer. Also dropped the per-directory count table, which carried a "do not work from this table" warning and went stale every session, and compressed the co-agent narrative to the diagnostic commands worth keeping. 280 -> 246. Recorded there too: the harness needs no main(), which matters because none of the remaining near-miss shell targets have one, so verify_shell_split.sh cannot check any of them. Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_011w2Pj4Ek3ZNMQVYcFHgXDY
This commit is contained in:
parent
0734e06ca5
commit
eef14d7bce
@ -52,6 +52,43 @@ the system, and diff the generated artifacts against a detached worktree at the
|
|||||||
pre-split commit. That comparison is what proved the fix — same file set, every
|
pre-split commit. That comparison is what proved the fix — same file set, every
|
||||||
text artifact byte-identical.
|
text artifact byte-identical.
|
||||||
|
|
||||||
|
### `trace_shell_split.sh` — the run, made routine
|
||||||
|
|
||||||
|
`meta/scripts/trace_shell_split.sh` (added 2026-08-16) does the stubbed run for
|
||||||
|
you. It shadows ~30 mutating binaries (`sudo`, `pacman`, `systemctl`, `adb`,
|
||||||
|
`nft`, `mount`, `mkinitcpio`, `reboot`, …) with `PATH` stubs that record each
|
||||||
|
call and exit 0, runs the script, and prints a diffable trace: exit status,
|
||||||
|
stubbed calls in order with arguments, stdout, stderr.
|
||||||
|
|
||||||
|
```bash
|
||||||
|
# baseline, in a detached worktree at the pre-split commit
|
||||||
|
meta/scripts/trace_shell_split.sh <script> --out /tmp/before.txt
|
||||||
|
# after the split
|
||||||
|
meta/scripts/trace_shell_split.sh <script> --out /tmp/after.txt
|
||||||
|
diff /tmp/before.txt /tmp/after.txt
|
||||||
|
```
|
||||||
|
|
||||||
|
It was verified against both bugs above: the `set -e` function tail shows up as
|
||||||
|
exit 1 with the later calls missing from the trace, and the nameref shows up as
|
||||||
|
`circular name reference` in stderr despite exit 0 and correct stdout. A run
|
||||||
|
against a fixture calling `systemctl` left the real unit untouched.
|
||||||
|
|
||||||
|
Two things it does not do:
|
||||||
|
|
||||||
|
- **Stubs always exit 0.** A script that branches on a real failure status
|
||||||
|
takes a path it would not take live. Hand-write a stub for those and say so
|
||||||
|
in the split's evidence file.
|
||||||
|
- **Anything not stubbed runs for real.** `git`, `curl`, `wget` and `makepkg`
|
||||||
|
are deliberately excluded from the default set, because many scripts read git
|
||||||
|
state harmlessly and a blanket stub would change what they see. Grep the
|
||||||
|
target for network and build verbs, then pass them:
|
||||||
|
`--stub git,curl,makepkg`.
|
||||||
|
|
||||||
|
Note it does not need a `main()`, which matters: none of the remaining
|
||||||
|
near-miss shell targets (`install_usage_monitoring.sh`,
|
||||||
|
`install_pacman_wrapper.sh`, `disk_cleanup_check.sh`, `fresh-install/main.sh`)
|
||||||
|
have one, so `verify_shell_split.sh` cannot check any of them.
|
||||||
|
|
||||||
Note the knock-on: renaming the nameref local hides the array from shellcheck
|
Note the knock-on: renaming the nameref local hides the array from shellcheck
|
||||||
(SC2034), because it is then only ever named as a bare word. Pass the name
|
(SC2034), because it is then only ever named as a bare word. Pass the name
|
||||||
through a variable (`CODE_FILES_MAP="LANG_CODE_FILES"`) so the reference is
|
through a variable (`CODE_FILES_MAP="LANG_CODE_FILES"`) so the reference is
|
||||||
|
|||||||
@ -22,25 +22,19 @@ files.
|
|||||||
|
|
||||||
`git log --oneline --grep='[Ss]plit'` is the authoritative list.
|
`git log --oneline --grep='[Ss]plit'` is the authoritative list.
|
||||||
|
|
||||||
## What is left (61)
|
## What is left
|
||||||
|
|
||||||
| Directory | Count |
|
60 files, plus this handoff whenever an edit pushes it over. A per-directory
|
||||||
| --------------------- | ------------- |
|
table used to live here and went stale every session; get the real list from:
|
||||||
| `linux_configuration` | 32 |
|
|
||||||
| `phone_focus_mode` | 11 (do LAST) |
|
|
||||||
| `focus_owner` | 5 |
|
|
||||||
| `kcd2_dice_solver` | 4 |
|
|
||||||
| `meta` | 3 |
|
|
||||||
| `bucket_catch` | 2 |
|
|
||||||
| `reverse_survivors` | 2 |
|
|
||||||
| `billsplit` | 1 (generated) |
|
|
||||||
|
|
||||||
**Do not work from this table.** Run:
|
|
||||||
|
|
||||||
```bash
|
```bash
|
||||||
bash ~/utils/scripts/check_file_length.sh --all
|
bash ~/utils/scripts/check_file_length.sh --all
|
||||||
```
|
```
|
||||||
|
|
||||||
|
Rough shape: `linux_configuration` dominates, `phone_focus_mode` (11) is last
|
||||||
|
by policy, then `focus_owner`, `kcd2_dice_solver`, `meta`, `bucket_catch`,
|
||||||
|
`reverse_survivors`.
|
||||||
|
|
||||||
### Deliberately over the cap — a NAMED blocker, not a line-count job
|
### Deliberately over the cap — a NAMED blocker, not a line-count job
|
||||||
|
|
||||||
Splitting one means fixing the blocker first. **Do not suppress to land one.**
|
Splitting one means fixing the blocker first. **Do not suppress to land one.**
|
||||||
@ -108,22 +102,10 @@ reachable by any static check:
|
|||||||
thing**: stub only what mutates the system, and diff the generated artifacts
|
thing**: stub only what mutates the system, and diff the generated artifacts
|
||||||
against a detached worktree at the pre-split commit.
|
against a detached worktree at the pre-split commit.
|
||||||
|
|
||||||
**`meta/scripts/trace_shell_split.sh` now does this** (added 2026-08-16). It
|
**`meta/scripts/trace_shell_split.sh` now does this** (added 2026-08-16) — it
|
||||||
shadows ~30 mutating binaries (`sudo`, `pacman`, `systemctl`, `adb`, `nft`,
|
runs the script with mutating binaries stubbed and prints a diffable trace, so
|
||||||
`mount`, `reboot`, …) with `PATH` stubs that record their calls and exit 0,
|
Decision 6 and Decision 12 can both hold. Usage, caveats and the proof it
|
||||||
runs the script, and prints a diffable trace of exit status + stubbed calls +
|
catches both bug classes are in `docs/shell-split-verification.md`.
|
||||||
stdout + stderr. Capture before and after a split, then diff.
|
|
||||||
|
|
||||||
Verified against both documented bug classes: it catches the `set -e`
|
|
||||||
function-tail abort (exit 1, later calls missing from the trace) and the
|
|
||||||
self-referencing nameref (exit 0 and correct stdout, but `circular name
|
|
||||||
reference` in stderr). Two caveats: stubs always exit 0, so a script branching
|
|
||||||
on a real failure status needs a hand-written stub; and a mutating command
|
|
||||||
**not** in `STUBBED_COMMANDS` runs for real — grep the script for mutating
|
|
||||||
verbs and extend the list before tracing something new.
|
|
||||||
|
|
||||||
This is what lets Decision 6 and "verify by running" coexist: the enforcement
|
|
||||||
and installer scripts are executed against stubs, never against the system.
|
|
||||||
|
|
||||||
### `--skip-install` is not permission to run a script
|
### `--skip-install` is not permission to run a script
|
||||||
|
|
||||||
@ -190,12 +172,12 @@ delete it.
|
|||||||
|
|
||||||
## Concurrency: check for another agent FIRST
|
## Concurrency: check for another agent FIRST
|
||||||
|
|
||||||
This session collided with a **second Claude** started by
|
A past session collided with a second Claude from
|
||||||
`~/.claude/scripts/claude-autoresume.sh`, running the same refactor in the same
|
`~/.claude/scripts/claude-autoresume.sh` on the same tree: shared git index, so
|
||||||
tree. Shared git index → its files landed in my staging area and it won a commit
|
its files landed in the other's staging area, it won a commit race, then hung
|
||||||
race. Then it **stalled for 17 minutes** and never recovered.
|
for 17 minutes on an API request with no read timeout. **Run this before
|
||||||
|
touching the index**, and `kill` anything live (it commits in-flight work on
|
||||||
Its signature, if it happens again:
|
the way out):
|
||||||
|
|
||||||
```bash
|
```bash
|
||||||
pgrep -af 'claude -p' | grep -v grep # autoresume runs `claude -p ... --continue`
|
pgrep -af 'claude -p' | grep -v grep # autoresume runs `claude -p ... --continue`
|
||||||
@ -204,25 +186,9 @@ pgrep -P <pid> -a # only MCP servers = no command running
|
|||||||
ss -tnp | grep <pid> # ESTAB to :443 with 0/0 queued = half-open
|
ss -tnp | grep <pid> # ESTAB to :443 with 0/0 queued = half-open
|
||||||
```
|
```
|
||||||
|
|
||||||
**Root cause:** a request to the API got no response, and there is no read
|
**If the tree looks suspiciously clean**, killing a process mid-`pre-commit`
|
||||||
timeout — so it waits forever in `epoll_wait`. Its log is 0 bytes because
|
stranded the changes it stashed. Recover them, do not retype them:
|
||||||
`claude -p` buffers all output until the run completes. `kill <pid>` is safe;
|
`ls -lt ~/.cache/pre-commit/patch* | head` then `git apply` the newest.
|
||||||
it committed its in-flight work on the way out.
|
|
||||||
|
|
||||||
**Before starting work:** run the `pgrep` above. If another agent is live in
|
|
||||||
this repo, deal with that before touching the index.
|
|
||||||
|
|
||||||
### If the tree looks suspiciously clean
|
|
||||||
|
|
||||||
Killing a process mid-`pre-commit` strands the unstaged changes it stashed.
|
|
||||||
They are recoverable — do **not** re-create them by hand:
|
|
||||||
|
|
||||||
```bash
|
|
||||||
ls -lt ~/.cache/pre-commit/patch* | head
|
|
||||||
git apply ~/.cache/pre-commit/patch<newest>
|
|
||||||
```
|
|
||||||
|
|
||||||
That is how `focus_owner/analysis_options.yaml` came back this session.
|
|
||||||
|
|
||||||
## Traps that still cost time
|
## Traps that still cost time
|
||||||
|
|
||||||
|
|||||||
Loading…
Reference in New Issue
Block a user