jake/SplitBit-Emulator
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

Flip says the true thing, and the checks that let it lie

Browse Source 
Two bugs, both in what the demo claimed rather than in the device.

The assembler has no string escapes, so the "\n" written in a literal
printed as a backslash and an n. A newline is a byte; Say.asm has always
written one as 0x0A 0x00 and this now writes it out of the console port.

And the line whose whole job was to still be there afterwards was wiped
out on the way back, because the program called osTakeScreen - which
restores the screen AS IT WAS BEFORE, so the tidy-up erased the one thing
the demo was pointing at. It did not need saving: nothing it touches is
the shell's. A program that damages nothing should not ask, and asking
anyway costs it the screen it was standing on.

Which turned out to be untrue as written, and that is the third thing.
Flip drew with a tile of its own, and the system copies the font back
over every tile at exit - so the filled screen went blank the moment the
program left, and the check that the system put the display back could
not tell a restored screen from an abandoned one. It passed with the
restore deleted. So did the check that a program can show the other
screen at all: a blank screen counts as one colour just as well as a
filled one does.

Now it fills with 0x0A, which is an asterisk in one of the reversed
colour schemes: paper is the colour and ink is black, so a whole screen
is drawn with NO TILE REDEFINED and it survives leaving. Both checks ask
for the commonest colour in the picture rather than counting colours or
naming a pixel - the font's only blank glyph is the space, whose
attribute nibble is nought, so a filled screen is always a pattern and
which pixel lands on paper depends on the character.

Both were re-broken afterwards and both failed this time.

Also cosmosFlip, a transcript test, which is what would have caught the
printed backslash in the first place.

Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01E2JrLzFvuFX9fgi1LDRjrW
This commit is contained in:
Anachronaut
2026-09-02 11:08:09 -04:00
co-authored by Claude Opus 5
parent 023362b05a
commit eee95ef0ce
15 changed files with 143 additions and 72 deletions
Download Patch File Download Diff File Expand all files Collapse all files
Tests/video.sh
+34 -22
View File
@@ -85,6 +85,21 @@ ASM
# bitmap pixel 5 are both address 0x0005. So every write below names the memory it is for,
# and there is deliberately no bare poke that picks by address - a helper that guessed would
# be right for the tiles and wrong for a picture, silently.
# The colour most of a picture is made of. Asked this way rather than by naming a pixel,
# because a filled screen is a PATTERN - the font has one blank glyph and it is the space,
# whose attribute nibble is nought - and which pixel lands on paper depends on the shape of
# whichever character was filled with.
commonest() {
# commonest <ppm>
python3 -c "
import collections, sys
d = open(sys.argv[1], 'rb').read()
px = d[d.index(b'255\n') + 4:]
counts = collections.Counter(px[o:o + 3] for o in range(0, len(px), 3))
print(counts.most_common(1)[0][0].hex())
" "$1" 2>/dev/null || echo none
}
pokeTo() {
# pokeTo <bank> <address> <byte>
printf ' INIA 0d%d\n OUTA 0xE3\n INIA 0x%02X\n OUTA 0xE4\n INIA 0x%02X\n OUTA 0xE5\n INIA 0x%02X\n OUTA 0xE9\n' \
@@ -1014,33 +1029,30 @@ python3 -c "open('$BUILD/flip.keys','wb').write(b'Flip\n' + b'\x00'*3000 + b' '
timeout 30 "$EMU" --fast --cycles 200000000 --keyboard "$BUILD/flip.keys" \
--screen "$BUILD/flip.ppm" --disk "$ROOT/Tests/build/disks/cosmos.img" \
--ram-disk 2048 "$BUILD/cosmos.bin" > "$BUILD/flip.out" 2>&1 || true
FLIPPED="$(python3 -c "
d = open('$BUILD/flip.ppm', 'rb').read()
px = d[d.index(b'255\n') + 4:]
print(len({px[o:o + 3] for o in range(0, len(px), 3)}))
" 2>/dev/null || echo 0)"
[ "$FLIPPED" = "1" ] \
&& result ok "a program shows the other screen" "the whole picture is the one it filled" \
|| result no "a program shows the other screen" "$FLIPPED colours, not a filled screen"
FLIPPED="$(commonest "$BUILD/flip.ppm")"
[ "$FLIPPED" = "50c050" ] \
&& result ok "a program shows the other screen" "green, which is the paper it filled with" \
|| result no "a program shows the other screen" "commonest colour $FLIPPED, not the fill"
# ---- And the system takes it back ----
#
# The shell's scrollback, its prompt and every line the person typed are in screen NOUGHT.
# A program that exited while showing screen one would hand back a shell drawing perfectly
# onto a screen nobody had ever written to - blank, and looking for all the world like the
# machine had lost everything. Compared cell by cell against the same session without Flip
# in it, which is the same way Grid's restore is checked.
python3 -c "open('$BUILD/flipbefore.keys','wb').write(b'dir\n' + b'Say a line to come back to\n' + b'\x00'*200)"
python3 -c "open('$BUILD/flipafter.keys','wb').write(b'dir\n' + b'Say a line to come back to\n' + b'Flip\n' + b'\x00'*3000 + b' ' + b'\x00'*3000 + b' ' + b'\x00'*3000)"
for phase in flipbefore flipafter; do
timeout 30 "$EMU" --fast --cycles 200000000 --keyboard "$BUILD/$phase.keys" \
--screen "$BUILD/$phase.ppm" --disk "$ROOT/Tests/build/disks/cosmos.img" \
--ram-disk 2048 "$BUILD/cosmos.bin" > "$BUILD/$phase.out" 2>&1 || true
done
BACK="$(python3 "$ROOT/Tests/samescreen.py" "$BUILD/flipbefore.ppm" "$BUILD/flipafter.ppm")"
[ "$BACK" = "yes" ] \
&& result ok "and the system puts the screen back" "screen nought, with what was on it" \
|| result no "and the system puts the screen back" "$BACK"
# onto a screen nobody had ever written to, and Flip does exit while flipped - deliberately,
# because a program that FAULTED while flipped could not put it back either.
#
# What says so is a corner of the screen with nothing on it. The map Flip filled is one tile
# and one attribute in every cell, so a screen still showing it is that colour EVERYWHERE; a
# screen nought that came back is black where nobody has printed. Checking a corner rather
# than comparing whole pictures, because Flip's own line is meant to survive on this one and
# an equality check would call that a difference.
python3 -c "open('$BUILD/flipafter.keys','wb').write(b'Say a line to come back to\n' + b'Flip\n' + b'\x00'*3000 + b' ' + b'\x00'*3000 + b' ' + b'\x00'*3000)"
timeout 30 "$EMU" --fast --cycles 200000000 --keyboard "$BUILD/flipafter.keys" \
--screen "$BUILD/flipafter.ppm" --disk "$ROOT/Tests/build/disks/cosmos.img" \
--ram-disk 2048 "$BUILD/cosmos.bin" > "$BUILD/flipafter.out" 2>&1 || true
[ "$(commonest "$BUILD/flipafter.ppm")" = "000000" ] \
&& result ok "and the system puts the screen back" "black again, and not the filled screen" \
|| result no "and the system puts the screen back" "commonest colour $(commonest "$BUILD/flipafter.ppm")"
# ---- Clearing puts the cursor back at the top ----
#