cbf1a48f

test: the e2e suite stops depending on a fast reap, a pager's height and a $TMPDIR without a trailing slash

a73x 2026-09-04 10:16

Commit message

test: the e2e suite stops depending on a fast reap, a pager's height and a $TMPDIR without a trailing slash

Three more assumptions that held on Linux and not on macOS 26.

The reap leg read the predecessor's stranded agent directory straight off
a bound socket. The reap is not a boot step — it happens inside the
successor's own makeDir, when its first session is born — so the check
raced the thing it was asserting, and lost on a Mac. It waits for the
successor's own directory to appear and reads the predecessor's then.

The pager leg spelled `less +G`, which says "put the end of the file on
screen" and leaves the top wherever the build's status line puts it: 178
under less 704, 179 under less 668. `+178g` names the top row instead, so
the wheel's 24 rows are asserted against a number the pager was told
rather than one it chose.

macOS sets $TMPDIR to a per-user directory ending in `/`, so every path
this suite spells came out with `//` in it, and the two groups that
compare such a path as a string — a layout leaf against a bound socket, a
hosts line against the daemon attached to — failed on a spelling. The
slash is stripped once, before the first of the 191 call sites.

Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01SakwJEwD9dXBoRP5kWbemW

test/e2e_01_boot.sh

Old		New
@@ -417,6 +417,21 @@ AGENTLEFT=$(find "$(dirname "$SOCK3")" -maxdepth 1 -name "mux-agent-$D3DEAD-*" \|
417	echo "e2e FAIL: a SIGKILLed daemon left $AGENTLEFT agent directories for pid $D3DEAD (want 1)"; exit 1; }	417	echo "e2e FAIL: a SIGKILLed daemon left $AGENTLEFT agent directories for pid $D3DEAD (want 1)"; exit 1; }
418	start_daemon "$SOCK3" "$OUT.d3b.d" "daemon did not rebind the stale socket" --shell /bin/sh	418	start_daemon "$SOCK3" "$OUT.d3b.d" "daemon did not rebind the stale socket" --shell /bin/sh
419	D3PID=$DPID	419	D3PID=$DPID
		420	# The reap is not a boot step: it happens inside the successor's own
		421	# `makeDir`, which runs when its FIRST session is born — here, when the
		422	# stranded client above redials. A bound socket says nothing about that,
		423	# so wait for the successor's own directory to appear and read the
		424	# predecessor's only then. Asserting straight off start_daemon read a
		425	# directory that was removed microseconds later, which is a race Linux
		426	# won every time and a Mac lost (2026-09-04).
		427	_i=0
		428	until [ -n "$(find "$(dirname "$SOCK3")" -maxdepth 1 -name "mux-agent-$D3PID-*")" ]; do
		429	_i=$((_i + 1)); [ "$_i" -lt $(( 100 * TIME_SCALE )) ] \|\| {
		430	echo "e2e FAIL: the restarted daemon never made an agent directory of its own,"
		431	echo " so nothing here has reaped yet and the check below would lie:"
		432	find "$(dirname "$SOCK3")" -maxdepth 1 -name 'mux-agent-*'; exit 1; }
		433	sleep 0.1
		434	done
420	AGENTLEFT=$(find "$(dirname "$SOCK3")" -maxdepth 1 -name "mux-agent-$D3DEAD-*" \| wc -l)	435	AGENTLEFT=$(find "$(dirname "$SOCK3")" -maxdepth 1 -name "mux-agent-$D3DEAD-*" \| wc -l)
421	[ "$AGENTLEFT" -eq 0 ] \|\| {	436	[ "$AGENTLEFT" -eq 0 ] \|\| {
422	echo "e2e FAIL: the restarted daemon left its SIGKILLed predecessor's agent directory:"	437	echo "e2e FAIL: the restarted daemon left its SIGKILLed predecessor's agent directory:"

test/e2e_08_mouse.sh

Old		New
@@ -199,19 +199,19 @@ ok "an application that asked for the mouse gets the wheel, and the client does
199	# consumed by the mouse filter and then dropped. Measured on this very	199	# consumed by the mouse filter and then dropped. Measured on this very
200	# scenario before the fix: four notches, grid unmoved.	200	# scenario before the fix: four notches, grid unmoved.
201	#	201	#
202	# `less +G` on 200 lines: the view starts at the END, so a wheel-UP is the	202	# `less +178g` on 200 lines: line 178 is put at the TOP of the screen, so a
203	# direction with somewhere to go. Eight notches at less's default of three	203	# wheel-UP has 177 rows to go and the leg knows where it started. Eight
204	# rows a notch is 24 rows, one screenful, so the top row moves back by 24	204	# notches at less's default of three rows a notch is 24 rows, one
205	# and the row that was at the top leaves the screen entirely — a needle and	205	# screenful, so the top moves to 154 and 178 leaves the screen entirely —
206	# its negative.	206	# a needle and its negative.
207	#	207	#
208	# The two row numbers are READ off the grid before the wheel rather than	208	# `+G` is what this leg used to spell, and it is the one thing here that
209	# written here, because how many of the 24 rows a pager spends on its own	209	# cannot be written down: it says "put the END of the file on screen", and
210	# status line is the pager's business and differs between builds: less 704	210	# where the top of that view lands depends on how many of the 24 rows the
211	# (Debian) starts this file at 178 and less 668 (macOS 26) at 179, and a	211	# build spends on its own status line. less 704 (Debian) left the top at
212	# leg that spelled 178 and 154 failed on the Mac by exactly one row while	212	# 178 and less 668 (macOS 26) at 179, so a leg that spelled 178 and 154
213	# the wheel worked perfectly. What this leg is about is the 24, not where	213	# failed on the Mac by exactly one row while the wheel worked perfectly.
214	# the pager chose to begin.	214	# `g` names the top row instead of deriving it, and both builds obey it.
215	#	215	#
216	# LESS is cleared in the script rather than trusted: an operator with	216	# LESS is cleared in the script rather than trusted: an operator with
217	# `LESS=--mouse` exported would have a pager that DOES ask for the mouse,	217	# `LESS=--mouse` exported would have a pager that DOES ask for the mouse,
@@ -222,34 +222,24 @@ cat > "$LESSSH" <<EOF
222	#!/bin/sh	222	#!/bin/sh
223	LESS=	223	LESS=
224	export LESS	224	export LESS
225	exec less +G $LESSDATA	225	exec less +178g $LESSDATA
226	EOF	226	EOF
227	chmod +x "$LESSSH"	227	chmod +x "$LESSSH"
228	start_daemon "$SOCK35" "$OUT.pgr.d" "pager daemon never bound" --shell "$LESSSH"	228	start_daemon "$SOCK35" "$OUT.pgr.d" "pager daemon never bound" --shell "$LESSSH"
229	D32PID=$DPID	229	D32PID=$DPID
230	# Not wait_grid: the needle is anchored (`-qx`, a row that is EXACTLY the	230	# Not wait_grid: the needle is anchored (`-qx`, a row that is EXACTLY one
231	# file's last line, so a half-painted `2` does not pass) and the dump is	231	# of the file's lines, so a half-painted `19` does not pass) and the dump
232	# trimmed to the last rows, because a pager's whole grid is 24 rows of	232	# is trimmed to the last rows, because a pager's whole grid is 24 rows of
233	# numbers that say nothing about why it stopped short.	233	# numbers that say nothing about why it stopped short. 190 rather than the
		234	# top row 178: it is well inside the page either build draws, so the needle
		235	# does not depend on the height question the comment above settles.
234	i=0	236	i=0
235	until "$MUX" d dump --sock "$SOCK35" \| grep -qx "200"; do	237	until "$MUX" d dump --sock "$SOCK35" \| grep -qx "190"; do
236	i=$((i+1)); [ "$i" -lt $(( 100 * TIME_SCALE )) ] \|\| {	238	i=$((i+1)); [ "$i" -lt $(( 100 * TIME_SCALE )) ] \|\| {
237	echo "e2e FAIL: pager never reached the end of the file:"	239	echo "e2e FAIL: pager never painted the page it was sent to:"
238	"$MUX" d dump --sock "$SOCK35" \| tail -3; exit 1; }	240	"$MUX" d dump --sock "$SOCK35" \| tail -3; exit 1; }
239	sleep 0.1	241	sleep 0.1
240	done	242	done
241	# The row the pager put at the top, read now and not written down: the
242	# daemon's session is 80x24 and the ptyclient below attaches at 80x24, so
243	# nothing between here and the wheel moves the view.
244	PGRTOP=$("$MUX" d dump --sock "$SOCK35" \| sed -n '1p')
245	case "$PGRTOP" in
246	''\|[!0-9])
247	echo "e2e FAIL: pager: the top row of the settled grid is not a line number: '$PGRTOP'"
248	"$MUX" d dump --sock "$SOCK35" \| head -3; exit 1 ;;
249	esac
250	[ "$PGRTOP" -gt 24 ] \|\| {
251	echo "e2e FAIL: pager: top row $PGRTOP leaves no room for a screenful of wheel"
252	"$MUX" d dump --sock "$SOCK35" \| head -3; exit 1; }
253	set +e	243	set +e
254	hostroom pager	244	hostroom pager
255	XDG_STATE_HOME="$HOSTROOM" timeout 40 "$PTYCLIENT" --cols 80 --rows 24 --out "$OUT.pgr" --err "$OUT.pgr.err" \	245	XDG_STATE_HOME="$HOSTROOM" timeout 40 "$PTYCLIENT" --cols 80 --rows 24 --out "$OUT.pgr" --err "$OUT.pgr.err" \
@@ -271,13 +261,12 @@ rc0 "pager: ptyclient leg exited $RC:" "$OUT.pgr.log"
271	"$MUX" d dump --sock "$SOCK35" > "$OUT.pgrcap" 2>&1	261	"$MUX" d dump --sock "$SOCK35" > "$OUT.pgrcap" 2>&1
272	# A whole line, because the pager's status line carries the file's PATH —	262	# A whole line, because the pager's status line carries the file's PATH —
273	# which holds this run's pid and could spell any short number.	263	# which holds this run's pid and could spell any short number.
274	PGRWANT=$(( PGRTOP - 24 ))	264	grep -qx "154" "$OUT.pgrcap" \|\| {
275	grep -qx "$PGRWANT" "$OUT.pgrcap" \|\| {	265	echo "e2e FAIL: pager: the wheel did not move the alternate screen back a screenful"
276	echo "e2e FAIL: pager: the wheel did not move the alternate screen back 24 rows"	266	echo " (top was 178, so row 154 should be on screen):"
277	echo " (top was $PGRTOP, so row $PGRWANT should be on screen):"
278	cat "$OUT.pgrcap"; exit 1; }	267	cat "$OUT.pgrcap"; exit 1; }
279	grep -qx "$PGRTOP" "$OUT.pgrcap" && {	268	grep -qx "178" "$OUT.pgrcap" && {
280	echo "e2e FAIL: pager: the old top row $PGRTOP is still on screen, so the view did not move a screenful:"	269	echo "e2e FAIL: pager: the old top row is still on screen, so the view did not move a screenful:"
281	cat "$OUT.pgrcap"; exit 1; }	270	cat "$OUT.pgrcap"; exit 1; }
282	# The pager never asked for the mouse, so it must never have SEEN a mouse	271	# The pager never asked for the mouse, so it must never have SEEN a mouse
283	# report: what reached it was arrow keys, which leave no text behind.	272	# report: what reached it was arrow keys, which leave no text behind.

test/e2e_lib.sh

Old		New
@@ -12,6 +12,26 @@
12	# What is NOT here: the scenario count. ok() counts, and the pin that	12	# What is NOT here: the scenario count. ok() counts, and the pin that
13	# gates the count is the runner's, at the bottom of test/e2e.sh — a group	13	# gates the count is the runner's, at the bottom of test/e2e.sh — a group
14	# file knows nothing about how many scenarios the suite has.	14	# file knows nothing about how many scenarios the suite has.
		15	# ---- $TMPDIR, trailing slash removed -----------------------------------
		16	# Every temporary path in this suite is spelled `${TMPDIR:-/tmp}/name`, and
		17	# a good few of them are then compared as STRINGS — a layout leaf against
		18	# the socket the daemon bound, a hosts line against the daemon the client
		19	# attached to. macOS sets $TMPDIR to a per-user directory ending in `/`, so
		20	# those paths came out with `//` in the middle: mux records the spelling it
		21	# was handed, the suite spells its expectation the same way, and the two
		22	# still differ when only one of them went through a normalizing step. Two
		23	# groups failed on exactly that. Stripped once here rather than at 191 call
		24	# sites, and before the first of them.
		25	while :; do
		26	case "${TMPDIR:-}" in
		27	*/) TMPDIR="${TMPDIR%/}" ;;
		28	*) break ;;
		29	esac
		30	done
		31	# An empty $TMPDIR is not a directory; leave it unset so `${TMPDIR:-/tmp}`
		32	# falls back the way every caller here expects.
		33	[ -n "${TMPDIR:-}" ] && export TMPDIR \|\| unset TMPDIR
		34
15	# ---- the cleanup registry ---------------------------------------------	35	# ---- the cleanup registry ---------------------------------------------
16	# Every process and every artifact a leg creates is REGISTERED where it is	36	# Every process and every artifact a leg creates is REGISTERED where it is
17	# created, and the EXIT trap walks the registers. Before this, cleanup()	37	# created, and the EXIT trap walks the registers. Before this, cleanup()