tests/deployment_scoping.bats:39 asserted [ -f scripts/.AGENTTRASH/chat-template.jinja.ik-llama-qwen-sharp ] so that the planning assertion "could not pass by the file having vanished". Both halves of that reasoning are wrong:
the planner's rule is a PATH rule — it plans nothing for that path wherever the bytes sit, so the existence guard protects no behavior;
the .AGENTTRASH tree is a purgeable hold, not an archive. #111 untracked it, so a fresh checkout or CI box has no such file and the suite goes red on bytes that carry no meaning.
Measured
fresh worktree off origin/main: deployment_scoping.bats:39fails (reproduced)
after the change, that file: 28 ok / 0 fail
./tests/run.sh on the branch: 789 ok, exit 0
The behavior assertion (the retired ik-llama.cpp chat template plans nothing, on both targets) is kept; only the disposable-existence half is cut.
## Why
`tests/deployment_scoping.bats:39` asserted `[ -f scripts/.AGENTTRASH/chat-template.jinja.ik-llama-qwen-sharp ]` so that the planning assertion "could not pass by the file having vanished". Both halves of that reasoning are wrong:
- the planner's rule is a **PATH** rule — it plans nothing for that path wherever the bytes sit, so the existence guard protects no behavior;
- the `.AGENTTRASH` tree is a **purgeable hold**, not an archive. #111 untracked it, so a fresh checkout or CI box has no such file and the suite goes red on bytes that carry no meaning.
## Measured
- fresh worktree off `origin/main`: `deployment_scoping.bats:39` **fails** (reproduced)
- after the change, that file: **28 ok / 0 fail**
- `./tests/run.sh` on the branch: **789 ok, exit 0**
The behavior assertion (the retired ik-llama.cpp chat template plans nothing, on both targets) is kept; only the disposable-existence half is cut.
---
cwd: `/home/blank/.omp/wt/wt-20261030-163847-6d1fc1f` · session: 01a18f2e-18b1-7573-bcd5-ac99636f8bcb
The seam asserted `[ -f scripts/.AGENTTRASH/chat-template.jinja.ik-llama-qwen-sharp ]`
so that the planning assertion "could not pass by the file having vanished". That
reasoning was wrong twice over: the planner's rule is a PATH rule, so it plans
nothing for that path no matter where the bytes sit, and the trash is a purgeable
hold — #111 untracked it, so a fresh worktree or CI checkout has no such file and
the suite went red on a byte that carries no meaning.
Measured: fresh worktree off origin/main, deployment_scoping.bats:39 fails. After
the change: that file 28 ok / 0 fail, ./tests/run.sh 789 ok exit 0.
dank
merged commit 0336f42e2a into main2026-10-31 19:30:35 -07:00
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.
Why
tests/deployment_scoping.bats:39asserted[ -f scripts/.AGENTTRASH/chat-template.jinja.ik-llama-qwen-sharp ]so that the planning assertion "could not pass by the file having vanished". Both halves of that reasoning are wrong:.AGENTTRASHtree is a purgeable hold, not an archive. #111 untracked it, so a fresh checkout or CI box has no such file and the suite goes red on bytes that carry no meaning.Measured
origin/main:deployment_scoping.bats:39fails (reproduced)./tests/run.shon the branch: 789 ok, exit 0The behavior assertion (the retired ik-llama.cpp chat template plans nothing, on both targets) is kept; only the disposable-existence half is cut.
cwd:
/home/blank/.omp/wt/wt-20261030-163847-6d1fc1f· session: 01a18f2e-18b1-7573-bcd5-ac99636f8bcb