35d86741

docs: record what the dedupe measured on contact

a73x 2026-09-01 19:13

Commit message

docs: record what the dedupe measured on contact

Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>

docs/superpowers/specs/2026-09-01-e2e-dedupe-and-vm-gate-design.md

Old		New
@@ -19,29 +19,34 @@ The retro record names the costs this design answers:
19		19
20	## Part 1 — dedupe (no behavior change)	20	## Part 1 — dedupe (no behavior change)
21		21
22	Three lib helpers replace three copied idioms. The scenario pins (105/38 in	22	As designed: three lib helpers. As measured on contact (2026-09-01): one
23	`test/e2e.sh`) do not move; that they do not move is the delivery check that	23	survived. The scenario pins (105/38 in `test/e2e.sh`) do not move; that
24	this part changed nothing.	24	they do not move is the delivery check that this part changed nothing.
25		25
26	1. `ssh_shim_rig DIR` in `e2e_lib.sh`: creates DIR, the three logs	26	1. `ssh_shim_rig DIR` — DROPPED. Reading all five shim bodies showed
27	(`DIR/pids`, `DIR/comms`, `DIR/args`), writes the option-loop head (today's	27	the common factor is already extracted: `ssh_shim_head` IS it, done
28	`ssh_shim_head`) plus the pid/comm/arg logging lines every rig repeats,	28	deliberately. The bodies differ in exactly what each group pins — 04 logs
29	registers `defer_rm`, and leaves the file open for the caller to append its	29	pid/comm/args, 09 remaps a runtime dir per fake box, 09's second shim
30	bespoke behavior before `chmod +x`. `04_handoff`, `09_hosts`, `15_askpass`	30	fakes unreachable-with-a-reason, 16 remaps a whole HOME — and 15 refuses
31	and `16_push` convert; their per-group behaviors (HRUN homes, fakebin/uname	31	the shared head on purpose, with a comment saying why (its subject is the
32	fronting) stay in the group files as appended bodies.	32	options the head discards). A rig helper would have served one caller and
33		33	blurred the other four.
34	2. `must "message" cmd args...`: runs the command with the rc captured	34
35	inside the helper, and on nonzero prints the message and the rc and exits 1.	35	2. `rc0 "message" [FILE...]` (renamed from the spec's `must`, reshaped
36	Converts only the sites whose failure block is a simple message — the	36	to read the `RC` the enclosing `set +e … RC=$? … set -e` capture already
37	multi-line diagnostic blocks keep their current shape, because the per-leg	37	holds — a wrapper taking `cmd args...` cannot carry the heredocs most
38	failure prose is load-bearing and `must` must never genericize it. Expect	38	sites feed ptyclient). Fails the suite printing the caller's own prose and
39	roughly half to two-thirds of the ~123 sites to convert; the true count is	39	the named capture files. It appends no "(rc N)" suffix: the messages carry
40	measured during implementation, not promised here.	40	`$RC` in bespoke positions, some with rc legends ("134 = panic"). 58 of
41		41	the 82 plain `-eq 0` sites converted; the ~24 that compare RC to specific
42	3. A shared detach tail for ptyclient scripts: one lib variable holding	42	values, kill something, or dump state on the way out keep their bespoke
43	`send \x1cd` + `waitexit`, interpolated into the ~57 heredocs that end a	43	form, and that boundary is the point — the helper covers the shape, never
44	scripted session the same way.	44	the judgment.
		45
		46	3. A shared detach tail — DROPPED. The heredocs are quote-protected so
		47	`$$` and `$?` reach the REMOTE shell literally; interpolating a shared
		48	tail means unquoting every script, which breaks every marker. Two clear
		49	lines per site beat that trade.
45		50
46	Conversion gate: full `make e2e` after each group file converts (the suite is	51	Conversion gate: full `make e2e` after each group file converts (the suite is
47	linear and stateful; `E2E_ONLY` cannot vouch for a group's neighbors), and	52	linear and stateful; `E2E_ONLY` cannot vouch for a group's neighbors), and