Let the console edit a line, and let a file be a keyboard

BACKSPACE REACHED THE SHELL. A terminal in line mode does not hand a program every
keystroke: it collects a line, rubs out a backspace, and delivers the finished thing at
Return. CosmOS has always relied on that, and behind a window there is no terminal to do it,
so the raw 0x08 went into the command buffer. Correcting a typo produced a line that looked
perfectly right on the screen and matched no command at all - "I do not know: help".

So the console does it, because behind a window the console IS the terminal. In key mode it
does not, and must not: a program in key mode asked for every keystroke as it happens.

CosmOS now asks for eighty columns at boot. Its own help text is seventy-four characters
wide, and dir, the monitor and the assembler's messages all assume room. The machine still
wakes up in the smaller mode, which is right for a machine - it is the system that knows
what shape of screen its own output needs, and a game that wants forty columns says so.

AND A FILE CAN BE A KEYBOARD, which is the part that matters beyond today. The console
behind a window is not the console behind a terminal, and until now the difference was
unreachable: it broke twice in two days and a person typing found it both times. --keyboard
installs the same hook a window does, so the same path runs, and the manifest has a column
for it. cosmosTyped types "halp", backs over it, arrives at "help", and requires the help to
come out. Verified by removing the rub-out, which loses the whole help text.

It does not test the window. Voyager's key queue and everything about presenting frames are
still out of reach. It tests the console, which is where the logic is.

Along the way: VOY_OBJS was missing from the dependency include, so voyager.o never rebuilt
when a header changed. EmulatorOptions grew a field, Voyager kept an object that disagreed
about the size of the struct, and smashed its stack on every run. A clean build hides it and
'make sanitize' cleans first, so that would never have found it either. Tests/voyager.sh did,
by failing all 115 tests that start the machine - which is the differential test earning its
keep on a bug that has nothing to do with what it was built to check.

Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01E2JrLzFvuFX9fgi1LDRjrW
This commit is contained in:
Anachronaut
2026-08-28 23:39:35 -04:00
co-authored by Claude Opus 5
parent 761c11a66b
commit 978aec4809
11 changed files with 214 additions and 17 deletions
+12
View File
@@ -51,6 +51,18 @@
#Program
boot:
; ---- The screen this system wants ----
;
; Eighty columns, because that is what CosmOS was written for: its own help text is
; seventy-four characters wide, and dir, the monitor and the assembler's messages all
; assume room. A machine wakes up in the smaller mode, which is right for a machine - it
; is the system that knows what shape of screen its own output needs.
;
; Harmless where there is no screen. Writing to a port nothing answers on does nothing,
; so this is one instruction wasted on a machine with a terminal instead.
INIA 0x01
OUTA 0x31
SETD.0 Banner
CALL printString
CALL newLine
+79 -12
View File
@@ -320,6 +320,61 @@ static void consoleDraw(uint8_t byte) {
// when one is typed, or -1 to say the window has gone.
static int (*inputHook)(int mayWait) = NULL;
// ---- Line editing, which the terminal used to do ----
//
// A TERMINAL IN LINE MODE DOES NOT HAND A PROGRAM EVERY KEYSTROKE. It collects a line,
// rubs out a backspace, and delivers the finished thing when Return is pressed. CosmOS has
// always relied on that, and behind a window there is no terminal to do it - so the raw
// backspace reached the shell, which put 0x08 in its command buffer and then could not find
// a command by that name. Correcting a typo made the line unrecognisable while looking
// perfectly right on screen.
//
// So the console does it, because behind a window the console IS the terminal. In key mode
// it does not: a program in key mode asked for every keystroke as it happens, which is the
// whole point of key mode.
#define CONSOLE_LINE_BYTES 256
static unsigned char consoleLine[CONSOLE_LINE_BYTES];
static int consoleLineLength = 0;
static int consoleLineAt = 0;
// Collects until Return, echoing as it goes, and leaves the line to be handed out a byte at
// a time. Returns 0 if the window closed while it was waiting.
static int consoleGatherLine(void) {
consoleLineAt = 0;
consoleLineLength = 0;
for (;;) {
const int got = inputHook(1);
if (got == CONSOLE_GONE) {
return 0;
}
if (got < 0) {
continue;
}
if (got == 0x08) {
// Nothing to rub out at the start of a line, and rubbing out past it would eat
// the prompt, which belongs to whoever printed it.
if (consoleLineLength > 0) {
consoleLineLength--;
consoleDraw(0x08);
}
continue;
}
if (got == '\n' || got == '\r') {
consoleLine[consoleLineLength++] = '\n';
consoleDraw('\n');
return 1;
}
// No room, or nothing a line is made of. A control byte that means something to a
// terminal means nothing here yet, and putting it in the line would only hand a
// program something it cannot use.
if (got < 0x20 || consoleLineLength >= CONSOLE_LINE_BYTES - 1) {
continue;
}
consoleLine[consoleLineLength++] = (unsigned char)got;
consoleDraw((uint8_t)got);
}
}
void consoleSetInputHook(int (*hook)(int mayWait)) {
inputHook = hook;
}
@@ -337,19 +392,21 @@ uint8_t consoleReadByte(void) {
}
consoleShowWhatIsWritten();
if (inputHook != NULL) {
for (;;) {
int got = inputHook(1);
if (got >= 0) {
// ---- The screen is the terminal now ----
//
// In line mode a terminal echoes what is typed and rubs out a backspace,
// and CosmOS has always relied on that. There is no terminal behind a
// window, so the display controller does it - which is exactly the job a
// video terminal's character generator had. In key mode nothing echoes,
// because a program in key mode is drawing its own screen.
if (!consoleKeyMode) {
consoleDraw((uint8_t)got);
// A line already gathered is handed out a byte at a time, which is what the
// program is asking for. Only when it runs out is another one collected.
if (consoleLineAt >= consoleLineLength && !consoleGatherLine()) {
consoleEnded = 1;
return 0xFF;
}
return consoleLine[consoleLineAt++];
}
for (;;) {
const int got = inputHook(1);
if (got >= 0) {
// Nothing echoes in key mode: a program that asked for every keystroke as it
// happens is drawing its own screen, and marks it did not make would be in
// the way.
return (uint8_t)got;
}
if (got == CONSOLE_GONE) {
@@ -408,7 +465,17 @@ static void consoleFetch(void) {
// the window happened to be focused. Asked without waiting, because a poll is a poll -
// the front end presents a frame when the console genuinely blocks, not when it looks.
if (inputHook != NULL) {
int got = inputHook(0);
if (!consoleKeyMode) {
// READY means there is a byte to be had, and in line mode there is one only
// while a gathered line is still being handed out. A poll must not take a key
// from under the gatherer, and half a line is not a line.
if (consoleLineAt < consoleLineLength) {
consolePushback = consoleLine[consoleLineAt++];
consoleAnnounce();
}
return;
}
const int got = inputHook(0);
if (got >= 0) {
consolePushback = got;
consoleAnnounce();
+39
View File
@@ -72,6 +72,32 @@ static void reportCycles(const CPURegisters *cpu, unsigned long cycleCount) {
}
// ---- A keyboard made of a file ----
//
// THE CONSOLE BEHIND A WINDOW IS NOT THE CONSOLE BEHIND A TERMINAL, and until this existed
// the difference was untestable. A terminal does the line editing; a window has none, so the
// console does it itself - gathering a line, rubbing out a backspace, handing it over only
// when Return arrives. That is real logic, it broke twice in two days, and both times it was
// found by a person typing rather than by anything here.
//
// So a file can be a keyboard. It installs the same hook a window does, which means the same
// path runs, and the suite can check what happens when a backspace arrives with nobody to
// interpret it. It does not test the window - Voyager's own key queue is still beyond reach
// - but it tests the console, which is where the logic is.
static FILE *keyboardFile = NULL;
static int keyboardHook(int mayWait) {
(void)mayWait; // There is no window to keep alive, so both questions are the same.
if (keyboardFile == NULL) {
return CONSOLE_GONE;
}
const int byte = fgetc(keyboardFile);
if (byte == EOF) {
return CONSOLE_GONE;
}
return byte & 0xFF;
}
uint8_t machineStart(Machine *m, const EmulatorOptions *options, const char *programFile) {
m->options = *options;
m->programFile = programFile;
@@ -115,6 +141,14 @@ uint8_t machineStart(Machine *m, const EmulatorOptions *options, const char *pro
if (m->options.debug) {
printRegisters(&m->cpu, Program, Data);
}
if (options->keyboard != NULL) {
keyboardFile = fopen(options->keyboard, "rb");
if (keyboardFile == NULL) {
fprintf(stderr, "Error: Couldn't read the keyboard file: %s\n", options->keyboard);
return MACHINE_ERROR;
}
consoleSetInputHook(keyboardHook);
}
setDiskLatency(m->options.diskCycles);
cycle_timer_init(&m->timer, CYCLE_RATE);
return MACHINE_OK;
@@ -220,6 +254,11 @@ void machineStop(Machine *m) {
if (m->options.screen != NULL) {
videoWriteImage(m->options.screen);
}
if (keyboardFile != NULL) {
consoleSetInputHook(NULL);
fclose(keyboardFile);
keyboardFile = NULL;
}
detachDisk();
}
+10 -1
View File
@@ -30,6 +30,10 @@ void printHelp(const char *programName) {
printf(" -S, --screen FILE Save a picture of the screen, as a PPM, when the machine\n");
printf(" stops. Works with or without a window, which is how the\n");
printf(" tests look at a screen on a host that has no display.\n");
printf(" -K, --keyboard FILE Feed the console from this file as though it were a\n");
printf(" keyboard rather than a terminal. Which means the console does\n");
printf(" its own line editing, the way it must when a window is open\n");
printf(" and there is no terminal behind it to do it.\n");
printf(" -h, --help Display this help message.\n");
}
@@ -42,6 +46,7 @@ uint8_t parseOptions(int argc, char *argv[], EmulatorOptions *options) {
{"write-protect", no_argument, 0, 'W'},
{"disk-cycles", required_argument, 0, 'L'},
{"screen", required_argument, 0, 'S'},
{"keyboard", required_argument, 0, 'K'},
{"help", no_argument, 0, 'h'},
{0, 0, 0, 0 }
};
@@ -55,9 +60,10 @@ uint8_t parseOptions(int argc, char *argv[], EmulatorOptions *options) {
options->writeProtect = 0;
options->diskCycles = 0;
options->screen = NULL;
options->keyboard = NULL;
// Parse options
while ((opt = getopt_long(argc, argv, "dc:fhD:WL:S:", long_options, &option_index)) != -1) {
while ((opt = getopt_long(argc, argv, "dc:fhD:WL:S:K:", long_options, &option_index)) != -1) {
switch (opt) {
case 'd':
options->debug = 1;
@@ -90,6 +96,9 @@ uint8_t parseOptions(int argc, char *argv[], EmulatorOptions *options) {
case 'S':
options->screen = optarg;
break;
case 'K':
options->keyboard = optarg;
break;
case 'h':
printHelp(argv[0]);
return OPTIONS_HELP;
+1
View File
@@ -23,6 +23,7 @@ typedef struct {
const char *disk; // Disk image to attach, or NULL for a machine with no disk.
uint8_t writeProtect; // Attach the disk read only, the way a tab on a floppy would.
const char *screen; // Where to save a picture of the screen when the machine stops.
const char *keyboard; // Feed the console from this file as a keyboard, not a terminal.
} EmulatorOptions;
uint8_t parseOptions(int argc, char *argv[], EmulatorOptions *options);
+16 -1
View File
@@ -78,7 +78,7 @@ from `make`, not from here.
### 1. Recorded output
`Tests/run.sh` assembles each program named in `Tests/manifest`, runs it, and compares
everything it printed against a file in `Tests/expected`. 165 tests, of which 103 run, 35
everything it printed against a file in `Tests/expected`. 166 tests, of which 104 run, 35
only assemble, 16 are expected to fail to assemble, and 11 boot from ROM with no image
given at all.
@@ -328,6 +328,21 @@ The `@N` form deserves a note. Every other test runs with the disk's answer avai
before the next instruction, which is the one condition under which failing to wait looks
exactly like working.
**keys** names a file in `Tests/input` to be fed to the console as a *keyboard* rather than
as standard input, and the difference between those is the whole reason the field exists.
Standard input reaches a console that believes a terminal is doing the line editing, which
is true when there is one: the terminal collects a line, rubs out a backspace, and hands
over the finished thing at Return. **Behind a window there is no terminal**, so the console
does that itself, and that is real logic which nothing could reach. It broke twice in two
days and a person typing found it both times - once as keys that never arrived, once as a
corrected line that reached the shell with the backspaces still in it, looking perfectly
right on screen and matching no command at all.
A keyboard file installs the same hook a window does, so the same path runs. It does not
test the window: Voyager's own key queue is still out of reach, and so is anything about
presenting frames. It tests the console, which is where the logic is.
## Fixture Disks:
`Tests/makedisks.sh` builds 26 images with SplitDisk before anything runs, into
+16
View File
@@ -0,0 +1,16 @@
CosmOS
> dir list what is on the disk
load <file> read a program off the disk
run [words] start what was loaded, and tell it those words
<name> [words] look where you are and then in /Apps, and start that
cd [path] go to a directory, or to the root with nothing after it
mkdir <path> make a directory
rmdir <path> remove an empty one
delete <file> take it off the disk
rename <file> <to> call it something else
monitor look at memory, change it, and jump into it
help this
exit stop, or leave the monitor if you are in it
> halted
Execution halted.
[exit 0]
+2
View File
@@ -0,0 +1,2 @@
halpmelpelp
exit
+16 -1
View File
@@ -4,7 +4,7 @@
# One test per line, fields separated by '|'. Blank lines and lines starting
# with '#' are ignored.
#
# name | source | mode | stdin | limit | disk
# name | source | mode | stdin | limit | disk | keys
#
# source is relative to Programs/. Everything assembles from there with
# Libraries/ on the include path, and the binary is written into Tests/build.
@@ -25,6 +25,13 @@
# nothing a test writes can be seen by the next one. Leave it off for a machine with no
# disk, which is most of them. A trailing :ro attaches it write protected.
#
# keys names a file in Tests/input to be fed to the console as a KEYBOARD rather than as
# standard input, and the difference is the point. Standard input reaches a console that
# believes a terminal is handling the line editing, which is true when one is. A keyboard
# reaches a console that knows it has to do the editing itself - gathering a line, rubbing
# out a backspace, handing it over only at Return - which is what happens behind a window,
# where there is no terminal to do any of it.
#
# A disk name with a directory in it, such as disks/sbfs.img, is one of the images that
# Tests/makedisks.sh builds with SplitDisk before the run. Those are used as they stand,
# so a test can read a filesystem written by the other implementation of the format.
@@ -442,6 +449,14 @@ wedgedImage | Boot/wedged.asm | assemble | -
# about a problem it will not let you fix. Settle is a program rather than a shell word:
# one job, reached through SWI, replaceable, and callable by anything that comes to call
# programs in sequence.
# ---- The console doing a terminal's job ----
#
# Behind a window nothing is handling the line editing, so the console does it. Typing "halp",
# backing over it, and arriving at "help" has to reach the shell as "help" - it used to arrive
# with the backspaces still in it, which made a corrected line unrecognisable while looking
# perfectly right on the screen.
cosmosTyped | CosmOS/Source/cosmos.asm | run | - | - | disks/cosmos.img | cosmosTyped.keys
cosmosSettle | CosmOS/Source/cosmos.asm | run | settle.in | 90000000 | disks/settle.img
# Saying so when there is nothing to settle is as much a part of it as doing it.
cosmosSettled | CosmOS/Source/cosmos.asm | run | settle.in | 90000000 | disks/settled.img
+17 -1
View File
@@ -171,14 +171,16 @@ uncolour() {
sed -i -E 's/\x1b\[[0-9;]*m//g' "$1"
}
while IFS='|' read -r name src mode stdin limit disk; do
while IFS='|' read -r name src mode stdin limit disk keys; do
name="$(trim "$name")"
[ -z "$name" ] && continue
case "$name" in \#*) continue ;; esac
src="$(trim "$src")"
mode="$(trim "$mode")"; stdin="$(trim "$stdin")"
limit="$(trim "$limit")"; disk="$(trim "$disk")"
keys="$(trim "${keys:-}")"
[ -z "$disk" ] && disk="-"
[ -z "$keys" ] && keys="-"
wanted "$name" || continue
@@ -231,6 +233,20 @@ while IFS='|' read -r name src mode stdin limit disk; do
# clock, and --cycles for programs that never halt on their own, which
# bounds them by cycle count rather than by wall clock.
EMUARGS=(--fast "${EMULATOR_EXTRA[@]}")
# ---- A file standing in for a keyboard ----
#
# Not the same as standard input, and that is the whole point of it. Input from a
# file reaches a console that believes a terminal is doing the line editing; a
# keyboard reaches one that knows it has to do the editing itself, which is what
# happens behind a window. The second had no test at all until it broke twice.
if [ "$keys" != "-" ]; then
if [ ! -f "$INPUT/$keys" ]; then
FAIL=$((FAIL + 1)); FAILED_NAMES+=("$name")
report "FAIL" "$name" "missing keyboard fixture $keys"
continue
fi
EMUARGS+=(--keyboard "$INPUT/$keys")
fi
[ "$limit" != "-" ] && EMUARGS+=(--cycles "$limit")
# A disk starts fresh for every run, so a test cannot pass because of what a
# previous one left lying on it. The emulator makes the image if it is
+6 -1
View File
@@ -167,7 +167,12 @@ $(OBJ_DIR)/%.o: $(SRC_DIR_ASM)/%.c
$(CC) $(CFLAGS) $(POSIXFLAGS) $(DEPFLAGS) -c $< -o $@
# Pull in the header dependencies written out by the compiler above.
-include $(EMU_OBJS:.o=.d) $(ASM_OBJS:.o=.d) $(DSK_OBJS:.o=.d) $(LINT_OBJS:.o=.d)
# VOY_OBJS IS IN THIS LIST FOR A REASON. It was not, and voyager.o therefore never rebuilt
# when a header changed - so when EmulatorOptions grew a field, Voyager kept an object that
# disagreed with everything else about how big the struct was, and smashed its stack on every
# run. A clean build hid it, which is why 'make sanitize' would never have found it either.
# Tests/voyager.sh did, by failing all 115 tests that start the machine.
-include $(EMU_OBJS:.o=.d) $(VOY_OBJS:.o=.d) $(ASM_OBJS:.o=.d) $(DSK_OBJS:.o=.d) $(LINT_OBJS:.o=.d)
# ---- The strict build the README promises ----
#