Preserve byte-verification logs for functional trials

Co-Authored-By: Pi/OpenAI <288921227+claudypoo@users.noreply.github.com>
This commit is contained in:
wassname
2026-09-10 13:42:18 +08:00
co-authored by Pi/OpenAI
parent ac6ef19e91
commit 5cda3d6b1d
4 changed files with 69 additions and 0 deletions
@@ -0,0 +1,16 @@
Verification log for hello.txt
Generated: 2026-09-10T04:22:36Z
Checked path: /tmp/goals-edxeth-trial-HSahSm/project/hello.txt
--- wc -c hello.txt ---
22 hello.txt
--- od -An -tx1 hello.txt ---
68 65 6c 6c 6f 20 66 72 6f 6d 20 74 68 65 20 77
6f 72 6b 65 72 0a
--- expected hex (reference) ---
68 65 6c 6c 6f 20 66 72 6f 6d 20 74 68 65 20 77 6f 72 6b 65 72 0a
--- byte count comparison ---
observed=22 expected=22
--- hex comparison ---
observed hex == expected hex
--- result ---
RESULT: PASS
@@ -0,0 +1,7 @@
Pi/OpenAI: Node assert.deepStrictEqual(actual, Buffer.from("hello from the worker\n")) passed
bytes=22
hex=68656c6c6f2066726f6d2074686520776f726b65720a
scheduler={
"jobs": [],
"version": 1
}
@@ -0,0 +1,40 @@
Parent Node byte assertions PASS: solo-ok newline; original hello unchanged
# Solo recovery test
## User-visible result
Create solo.txt containing exactly `solo-ok\n` (8 UTF-8 bytes) and save actual byte-verification output in evidence/solo.log.
## Preferences
- Explicit solo mode: main session implements and edits; completion is self-verification, not an independent review.
- No other worker, installs, commits, network requests or unrelated file changes. Preserve hello.txt and evidence/hello.log.
## Goals
- [x] goal: Produce the exact solo.txt bytes and saved verification log
- subtle failure mode: main session reports success without writing the file, or labels self-verification independent review.
- discriminator: actual file is the 8 bytes `solo-ok\n`; saved log records a successful byte comparison; completion log identifies solo self-verification.
- evidence (satisfied): `solo.txt` = 8 bytes, `od -An -tx1 -v` reads exactly `73 6f 6c 6f 2d 6f 6b 0a` (= `solo-ok\n`); `evidence/solo.log` (441 bytes, nonempty) records real `wc -c` (8) and `od` output plus programmatic byte-count and hex comparisons against `736f6c6f2d6f6b0a` and `RESULT: PASS`; verified by the main session itself (solo self-verification, not independent review)
## Log
- Solo self-verification: "Produce the exact solo.txt bytes and saved verification log"; "Solo-mode self-verification (main session, not independent review). Inspected actual artifacts on disk: solo.txt exists at 8 bytes; `od -An -tx1 -v solo.txt` outputs exactly `73 6f 6c 6f 2d 6f 6b 0a`, i.e. UTF-8 `solo-ok\\n`. evidence/solo.log exists (441 bytes, nonempty) and records the real `wc -c` result (8), the real `od` dump, the expected-hex reference, programmatic byte-count and hex comparisons (observed=8 expected=8; actual hex 736f6c6f2d6f6b0a == expected hex), and `RESULT: PASS`. Log content matches the on-disk file byte-for-byte. Constraints held: no installs, no commits, no network calls; hello.txt and evidence/hello.log preserved untouched; no unrelated changes. Scheduler: list shows no jobs, so no job named goals-01a0898a-3dde-7015-b029-ced69dc7b553 exists to remove; nothing added/enabled/recreated. Goal satisfied."; evidence ["/tmp/goals-edxeth-trial-HSahSm/project/solo.txt","/tmp/goals-edxeth-trial-HSahSm/project/evidence/solo.log"]
Prepared by Pi/OpenAI as an isolated functional-test plan.
- Ran goal (main session, solo mode): wrote `solo.txt` via `printf 'solo-ok\n'`, then generated `evidence/solo.log` from real command output with a programmatic PASS/FAIL byte comparison; re-checked `wc -c` = 8 and hex `73 6f 6c 6f 2d 6f 6b 0a` directly. This is solo self-verification by the main session, not independent review. Constraints: no installs/commits/network; `hello.txt` and `evidence/hello.log` preserved. Goal complete.
Byte-verification log for solo.txt (solo recovery test)
Time: 2026-09-10T04:52:22Z
--- wc -c solo.txt ---
8 solo.txt
--- od -An -tx1 -v solo.txt ---
73 6f 6c 6f 2d 6f 6b 0a
--- expected hex (8 bytes: solo-ok\n) ---
73 6f 6c 6f 2d 6f 6b 0a
--- byte count comparison ---
observed=8 expected=8
--- hex comparison (actual vs expected) ---
actual=736f6c6f2d6f6b0a
expected=736f6c6f2d6f6b0a
actual hex == expected hex
--- result ---
RESULT: PASS
@@ -0,0 +1,6 @@
Parent independent byte assertion PASS: verified + LF, 9 bytes. Saved verification nonempty with PASS. Scheduler jobs empty.
== byte verification of verified.txt ==
9 verified.txt
76 65 72 69 66 69 65 64 0a
PASS: 9 bytes: verified + LF