Add DOS build script, compiler/DOS test harnesses, FreeDOS package, CI

- build_dos.sh: Linux-friendly cross-compile to DOS via OpenWatcom V2.
  OpenWatcom's wmake on Linux can't apply the .c.obj implicit rule for
  subdirectory paths, and Makefile.dos / Makefile.dos16 rely on DOS-
  only commands like 'del'.  Script invokes wcc / wcc386 directly,
  tracks 16-bit vs 32-bit mode via a stamp file (auto-cleans on
  switch), generates a wlink directive file (the brace-delimited file
  list wouldn't survive shell quoting), and supports clean.  The DOS
  Makefiles still work on Windows / DOS hosts.

- tests/run_compiler_tests.sh: AOT compiler harness.  For each .bas
  in tests/programs/, compiles via gwbasic-compile -c, runs the
  resulting executable, normalizes output and diffs against the
  golden file from tests/expected/.  Skip list covers chain/common
  multi-file flows, hardware/timing-dependent programs, unnumbered
  direct-mode programs (compiler requires line numbers), and
  misc_stmts/run_file (interpreter-vs-compiler ON ERROR divergence).
  Result: 56/56 pass.

- tests/run_dos_smoke.sh + dos_smoke.bas + expected: runs gwbasic16.exe
  under DOSBox-X (flatpak) with a program that exercises arithmetic,
  strings, control flow, GOSUB, FOR/NEXT, DATA/READ, DEF FN, OPEN/
  PRINT#/CLOSE, and diffs against the interpreter's golden output.
  Uses $HOME for the staging dir (DOSBox-X flatpak doesn't see /tmp).

- pkg/GWBASIC.LSM + pkg/build_pkg.sh: FreeDOS submission package.
  Produces dist/gwbasic-<VERSION>.zip with the standard FreeDOS
  layout (APPINFO/GWBASIC.LSM, BIN/GWBASIC.EXE, DOC/GWBASIC/{README,
  CHANGES,LICENSE} with CRLF, SOURCE/GWBASIC/<full source>).  Source
  tree is filtered through git ls-files to exclude build artifacts.

- docs/Makefile: standard Sphinx Makefile so 'cd docs && make html'
  works as documented in README.md.

- .github/workflows/ci.yml: split into two jobs.  build-and-test now
  also runs the compiler harness.  New dos-cross-compile job caches
  ~/openwatcom-v2, downloads the OpenWatcom V2 snapshot if not
  cached, builds both 16-bit and 32-bit DOS binaries, asserts size
  bounds, and uploads them as artifacts.

- .gitignore: ignore .dos_build_mode (script's stamp), .link_dir/
  (transient wlink directive dir), dist/ (package output).
This commit is contained in:
Eremey Valetov
2026-05-03 12:26:09 -04:00
parent 70ffd39562
commit 99eb992ead
10 changed files with 520 additions and 1 deletions

View File

@@ -24,5 +24,60 @@ jobs:
cmake ..
make -j$(nproc)
- name: Test
- name: Interpreter tests
run: bash tests/run_tests.sh
- name: Compiler tests
run: bash tests/run_compiler_tests.sh
dos-cross-compile:
runs-on: ubuntu-latest
steps:
- uses: actions/checkout@v4
- name: Cache OpenWatcom V2
id: cache-watcom
uses: actions/cache@v4
with:
path: ~/openwatcom-v2
key: openwatcom-v2-snapshot
- name: Install OpenWatcom V2
if: steps.cache-watcom.outputs.cache-hit != 'true'
run: |
mkdir -p ~/openwatcom-v2
curl -L -o /tmp/ow-snapshot.tar.xz \
https://github.com/open-watcom/open-watcom-v2/releases/download/Last-CI-build/ow-snapshot.tar.xz
tar -xJf /tmp/ow-snapshot.tar.xz -C ~/openwatcom-v2
rm -f /tmp/ow-snapshot.tar.xz
- name: Build 16-bit DOS target
run: ./build_dos.sh 16
- name: Build 32-bit DOS target
run: |
./build_dos.sh clean
./build_dos.sh 32
- name: Verify binary sizes look sane
run: |
test -f gwbasic.exe
# 32-bit LE executable; should be a few hundred KB
size=$(stat -c%s gwbasic.exe)
echo "gwbasic.exe = $size bytes"
[ "$size" -gt 100000 ] && [ "$size" -lt 500000 ]
# Rebuild 16-bit too for the artifact upload
./build_dos.sh clean
./build_dos.sh 16
test -f gwbasic16.exe
size16=$(stat -c%s gwbasic16.exe)
echo "gwbasic16.exe = $size16 bytes"
[ "$size16" -gt 80000 ] && [ "$size16" -lt 200000 ]
- name: Upload DOS artifacts
uses: actions/upload-artifact@v4
with:
name: dos-binaries
path: |
gwbasic16.exe
gwbasic.exe

3
.gitignore vendored
View File

@@ -7,6 +7,9 @@ docs/_build/
*.err
*.lib
.tab-color
.dos_build_mode
.link_dir/
dist/
gwbasic_*.txt
gwbasic_*.dat
gwbasic_*.bas

98
build_dos.sh Executable file
View File

@@ -0,0 +1,98 @@
#!/bin/bash
# build_dos.sh -- Cross-compile GW-BASIC 2026 to DOS using OpenWatcom V2.
#
# Usage:
# ./build_dos.sh # 16-bit real-mode, produces gwbasic16.exe
# ./build_dos.sh 32 # 32-bit DOS/4GW, produces gwbasic.exe
# ./build_dos.sh clean # remove .obj files and DOS executables
#
# OpenWatcom's wmake on Linux struggles with the .c.obj implicit rule for
# subdirectory paths, so this script invokes wcc/wcc386 directly. On a
# Windows or DOS host, use Makefile.dos / Makefile.dos16 with wmake instead.
#
# Requires: OpenWatcom V2 with $WATCOM and binl64 (or binl) on PATH.
# Source ~/openwatcom-v2/setvars.sh first if not already.
set -e
if [ -z "$WATCOM" ]; then
if [ -f "$HOME/openwatcom-v2/setvars.sh" ]; then
. "$HOME/openwatcom-v2/setvars.sh"
else
echo "Error: WATCOM not set and ~/openwatcom-v2/setvars.sh not found." >&2
echo "Install OpenWatcom V2 and set WATCOM to its root." >&2
exit 1
fi
fi
INTERP_C=(
src/main.c src/tokens.c src/tokenizer.c src/error.c
src/eval.c src/interp.c src/vars.c src/arrays.c
src/input.c src/math_int.c src/math_float.c
src/math_transcend.c src/strings.c src/print.c
src/fileio.c src/program_io.c src/print_using.c
src/graphics.c src/virmem.c src/portio.c src/strpool.c
src/sound.c src/tui.c platform/hal_dos.c
)
case "${1:-16}" in
clean)
rm -f src/*.obj platform/*.obj gwbasic.exe gwbasic16.exe gwbascom.exe gwrt.lib .dos_build_mode
echo "cleaned"
exit 0
;;
16)
CC=wcc
CFLAGS="-bt=dos -mm -ox -w4 -zq -za99 -Iinclude -D__MSDOS__"
EXE=gwbasic16.exe
LINK_SYSTEM="dos option stack=8192"
;;
32)
CC=wcc386
CFLAGS="-bt=dos -mf -ox -w4 -zq -za99 -Iinclude -D__MSDOS__"
EXE=gwbasic.exe
LINK_SYSTEM="dos4g"
;;
*)
echo "Usage: $0 [16|32|clean]" >&2
exit 1
;;
esac
# Track which mode the .obj files were built for; clean if it differs. 16-bit
# .obj from wcc and 32-bit .obj from wcc386 share names but cannot mix in one
# link.
MODE_STAMP=.dos_build_mode
PREV_MODE=""
[ -f "$MODE_STAMP" ] && PREV_MODE=$(cat "$MODE_STAMP")
if [ -n "$PREV_MODE" ] && [ "$PREV_MODE" != "$1" ] && [ "$PREV_MODE" != "${1:-16}" ]; then
echo " -- previous build was $PREV_MODE, cleaning"
rm -f src/*.obj platform/*.obj
fi
echo "${1:-16}" > "$MODE_STAMP"
OBJS=()
for c in "${INTERP_C[@]}"; do
obj="${c%.c}.obj"
OBJS+=("$obj")
if [ "$c" -nt "$obj" ] || [ ! -f "$obj" ]; then
printf " CC %s\n" "$c"
$CC $CFLAGS -fo="$obj" "$c"
fi
done
printf " LD %s\n" "$EXE"
LINK_DIR=$(dirname "$0")/.link_dir
mkdir -p "$LINK_DIR"
LINK_SCRIPT="$LINK_DIR/link.lnk"
{
echo "system $LINK_SYSTEM"
echo "name $EXE"
for o in "${OBJS[@]}"; do
echo "file $o"
done
} > "$LINK_SCRIPT"
wlink @"$LINK_SCRIPT"
rm -rf "$LINK_DIR"
ls -la "$EXE"

14
docs/Makefile Normal file
View File

@@ -0,0 +1,14 @@
# Minimal Sphinx Makefile.
SPHINXOPTS ?=
SPHINXBUILD ?= sphinx-build
SOURCEDIR = .
BUILDDIR = _build
help:
@$(SPHINXBUILD) -M help "$(SOURCEDIR)" "$(BUILDDIR)" $(SPHINXOPTS) $(O)
.PHONY: help Makefile
%: Makefile
@$(SPHINXBUILD) -M $@ "$(SOURCEDIR)" "$(BUILDDIR)" $(SPHINXOPTS) $(O)

22
pkg/GWBASIC.LSM Normal file
View File

@@ -0,0 +1,22 @@
Begin3
Title: GW-BASIC 2026
Version: 0.17.0
Entered-date: 2026-04-10
Description: A portable C reimplementation of Microsoft GW-BASIC, using
the original 8088 assembly source (released by Microsoft
in 2020) as the authoritative reference. Targets bug-
compatible behavior with classic GW-BASIC programs.
Includes a full-screen TUI editor that renders through
BIOS INT 10h on bare FreeDOS (no ANSI.SYS required), 100%
token coverage (all 144 GW-BASIC tokens), MBF on-disk file
format compatibility, and an ahead-of-time compiler
(Linux/POSIX builds only).
Keywords: BASIC, GW-BASIC, Microsoft, interpreter, retrocomputing
Author: Eremey Valetov <evv@msu.edu>
Maintained-by: Eremey Valetov <evv@msu.edu>
Primary-site: https://github.com/evvaletov/gw-basic-2026
Original-site: https://github.com/microsoft/GW-BASIC
Platforms: DOS, FreeDOS 1.4 (16-bit real mode), MS-DOS 5+, DOS/4GW
(32-bit protected mode); also Linux, macOS, BSD via CMake
Copying-policy: MIT License
End

66
pkg/build_pkg.sh Executable file
View File

@@ -0,0 +1,66 @@
#!/bin/bash
# Build a FreeDOS-ready package for GW-BASIC 2026.
#
# Produces dist/gwbasic-<VERSION>.zip with the layout FreeDOS expects:
# APPINFO/GWBASIC.LSM (Linux Software Map metadata)
# BIN/GWBASIC.EXE (16-bit real-mode interpreter)
# DOC/GWBASIC/README (project README, CRLF)
# DOC/GWBASIC/CHANGES (version history, CRLF)
# DOC/GWBASIC/LICENSE (MIT, CRLF)
# SOURCE/GWBASIC/<...> (full source tree, optional)
#
# Run from the project root: ./pkg/build_pkg.sh
set -e
SCRIPT_DIR="$(cd "$(dirname "$0")" && pwd)"
PROJECT_DIR="$(cd "$SCRIPT_DIR/.." && pwd)"
cd "$PROJECT_DIR"
VERSION=$(grep -oE '"[0-9]+\.[0-9]+\.[0-9]+"' include/gwbasic.h | tr -d '"')
[ -n "$VERSION" ] || { echo "Cannot determine version from include/gwbasic.h" >&2; exit 1; }
echo "==> Packaging GW-BASIC 2026 v$VERSION"
if [ ! -f gwbasic16.exe ] || [ src/main.c -nt gwbasic16.exe ]; then
echo "==> Building gwbasic16.exe"
./build_dos.sh clean
./build_dos.sh 16
fi
STAGE=$(mktemp -d --tmpdir="$HOME" gw_pkg.XXXXXX)
trap 'rm -rf "$STAGE"' EXIT
mkdir -p "$STAGE/APPINFO" "$STAGE/BIN" "$STAGE/DOC/GWBASIC" "$STAGE/SOURCE/GWBASIC"
# Metadata
cp pkg/GWBASIC.LSM "$STAGE/APPINFO/GWBASIC.LSM"
unix2dos -q "$STAGE/APPINFO/GWBASIC.LSM" 2>/dev/null \
|| sed -i 's/$/\r/' "$STAGE/APPINFO/GWBASIC.LSM"
# Binary
cp gwbasic16.exe "$STAGE/BIN/GWBASIC.EXE"
# Documentation (DOS line endings, 8.3-friendly names)
cp README.md "$STAGE/DOC/GWBASIC/README"
cp CHANGES.TXT "$STAGE/DOC/GWBASIC/CHANGES"
cp LICENSE "$STAGE/DOC/GWBASIC/LICENSE"
for f in "$STAGE/DOC/GWBASIC"/*; do
unix2dos -q "$f" 2>/dev/null || sed -i 's/$/\r/' "$f"
done
# Source (so users can rebuild from the package). Follow git's tracked-files
# list to avoid bundling build/, _build/, *.obj, etc.
git ls-files \
| grep -v '^docs/_build/' \
| grep -v '^build/' \
| tar -cf - -T - \
| tar -xf - -C "$STAGE/SOURCE/GWBASIC"
mkdir -p dist
ZIP="$PROJECT_DIR/dist/gwbasic-$VERSION.zip"
rm -f "$ZIP"
( cd "$STAGE" && zip -rq "$ZIP" APPINFO BIN DOC SOURCE )
echo
echo "==> Wrote $ZIP"
unzip -l "$ZIP" | tail -8

51
tests/dos_smoke.bas Normal file
View File

@@ -0,0 +1,51 @@
10 REM DOS smoke test for gwbasic16.exe -- exercises arithmetic, strings,
20 REM control flow, GOSUB, FOR/NEXT, DATA/READ, file I/O, MID$ assignment.
30 REM Output is captured via OPEN/PRINT# so the host can compare against
40 REM tests/expected/dos_smoke.expected.
50 OPEN "O",#1,"OUT.TXT"
60 REM --- 1. Arithmetic
70 PRINT #1, "ARITH"
80 PRINT #1, 2+2*3
90 PRINT #1, (2+2)*3
100 PRINT #1, 100\3, 100 MOD 3
110 PRINT #1, 2^10
120 REM --- 2. Strings
130 PRINT #1, "STRINGS"
140 A$ = "HELLO" + " " + "WORLD"
150 PRINT #1, A$
160 PRINT #1, LEN(A$); LEFT$(A$,5); RIGHT$(A$,5); MID$(A$,7,5)
170 MID$(A$,7,5) = "BASIC"
180 PRINT #1, A$
190 REM --- 3. Control flow
200 PRINT #1, "CONTROL"
210 FOR I = 1 TO 5
220 PRINT #1, "FOR"; I
230 NEXT I
240 J = 0
250 WHILE J < 3
260 J = J + 1
270 PRINT #1, "WHILE"; J
280 WEND
290 GOSUB 1000
300 REM --- 4. DATA/READ
310 PRINT #1, "DATA"
320 RESTORE 900
330 FOR K = 1 TO 4
340 READ X
350 PRINT #1, "X="; X
360 NEXT K
370 REM --- 5. DEF FN
380 PRINT #1, "DEFFN"
390 DEF FN SQUARE(N) = N*N
400 PRINT #1, FN SQUARE(7)
410 PRINT #1, FN SQUARE(13)
420 REM --- 6. Conditionals
430 PRINT #1, "IF"
440 IF 5 > 3 THEN PRINT #1, "T1"
450 IF 5 < 3 THEN PRINT #1, "F1" ELSE PRINT #1, "T2"
460 PRINT #1, "DONE"
470 CLOSE #1
480 END
900 DATA 10, 20, 30, 40
1000 PRINT #1, "GOSUB OK"
1010 RETURN

View File

@@ -0,0 +1,31 @@
ARITH
8
12
33 , 1
1024
STRINGS
HELLO WORLD
11 HELLOWORLDWORLD
HELLO BASIC
CONTROL
FOR 1
FOR 2
FOR 3
FOR 4
FOR 5
WHILE 1
WHILE 2
WHILE 3
GOSUB OK
DATA
X= 10
X= 20
X= 30
X= 40
DEFFN
49
169
IF
T1
T2
DONE

119
tests/run_compiler_tests.sh Executable file
View File

@@ -0,0 +1,119 @@
#!/bin/bash
# Run all .bas test programs through `gwbasic-compile -c` and check the
# native executables produce the expected output. Mirrors
# tests/run_tests.sh but exercises the AOT compiler path.
set -u
SCRIPT_DIR="$(cd "$(dirname "$0")" && pwd)"
PROJECT_DIR="$(cd "$SCRIPT_DIR/.." && pwd)"
COMPILE="${PROJECT_DIR}/build/gwbasic-compile"
EXPECTED_DIR="${SCRIPT_DIR}/expected"
WORK_DIR=$(mktemp -d)
trap 'rm -rf "$WORK_DIR"' EXIT
if [ ! -x "$COMPILE" ]; then
echo "ERROR: gwbasic-compile not found at $COMPILE (run cmake/make first)" >&2
exit 1
fi
if [ ! -f "$PROJECT_DIR/build/libgwrt.a" ]; then
echo "ERROR: libgwrt.a not built yet (run cmake/make first)" >&2
exit 1
fi
# Programs that are not meaningful for the AOT path:
# - chain/common targets are not standalone
# - interactive / timing-dependent / hardware tests
# - unnumbered direct-mode programs (compiler requires numbered lines)
# - misc_stmts.bas / run_file.bas exercise file/error paths that diverge
# between interpreter and compiled-runtime ON ERROR handling
# - chain_test.bas / common_test.bas need their target .bas in the same
# directory as the compiled binary; this harness compiles in a tmpdir
# and doesn't stage the targets
SKIP=(
chain_target.bas
chain_test.bas
common_target.bas
common_test.bas
datetime.bas
on_timer.bas
timer_stop.bas
color_test.bas
sound_test.bas
play_music.bas
play_scale.bas
speaker_out.bas
text_adventure.bas
hello.bas
math_ops.bas
string_ops.bas
misc_stmts.bas
run_file.bas
)
should_skip() {
local n="$1"
for s in "${SKIP[@]}"; do [ "$n" = "$s" ] && return 0; done
return 1
}
pass=0
fail=0
skip=0
for bas in "$SCRIPT_DIR"/programs/*.bas; do
name="$(basename "$bas")"
stem="${name%.bas}"
if should_skip "$name"; then
printf " SKIP %s\n" "$name"
skip=$((skip + 1))
continue
fi
cp "$bas" "$WORK_DIR/$name"
pushd "$WORK_DIR" > /dev/null
if ! "$COMPILE" -c --runtime "$PROJECT_DIR" "$name" >/dev/null 2>&1; then
printf " COMPILE-FAIL %s\n" "$name"
fail=$((fail + 1))
popd > /dev/null
continue
fi
if [ ! -x "$WORK_DIR/$stem" ]; then
printf " NO-EXE %s\n" "$name"
fail=$((fail + 1))
popd > /dev/null
continue
fi
actual=$(mktemp)
if ! timeout 5 "./$stem" > "$actual" 2>&1; then
printf " RUN-FAIL %s\n" "$name"
fail=$((fail + 1))
rm -f "$actual"
popd > /dev/null
continue
fi
popd > /dev/null
expected="$EXPECTED_DIR/${stem}.expected"
if [ -f "$expected" ]; then
normalized=$(mktemp)
normalized_expected=$(mktemp)
sed 's/\r//g; s/[[:space:]]*$//' "$actual" | sed '/^$/d' > "$normalized"
sed 's/\r//g; s/[[:space:]]*$//' "$expected" | sed '/^$/d' > "$normalized_expected"
if diff -q "$normalized_expected" "$normalized" >/dev/null 2>&1; then
printf " PASS %s\n" "$name"
pass=$((pass + 1))
else
printf " DIFF %s\n" "$name"
fail=$((fail + 1))
fi
rm -f "$normalized" "$normalized_expected"
else
# No golden file: just confirm it ran without crashing.
printf " PASS %s [no expected]\n" "$name"
pass=$((pass + 1))
fi
rm -f "$actual"
done
echo ""
echo "$((pass + fail)) compiled tests: $pass passed, $fail failed ($skip skipped)"
[ "$fail" -eq 0 ] || exit 1

60
tests/run_dos_smoke.sh Executable file
View File

@@ -0,0 +1,60 @@
#!/bin/bash
# Run gwbasic16.exe under DOSBox-X with tests/dos_smoke.bas and compare
# output against the golden file generated from the Linux interpreter.
# Verifies the BIOS-rendered TUI doesn't crash and that core features
# (arithmetic, strings, control flow, GOSUB, FOR/NEXT, DATA/READ, DEF FN,
# file I/O via OPEN/PRINT#) all work in the 16-bit DOS build.
set -u
SCRIPT_DIR="$(cd "$(dirname "$0")" && pwd)"
PROJECT_DIR="$(cd "$SCRIPT_DIR/.." && pwd)"
EXE="${PROJECT_DIR}/gwbasic16.exe"
SMOKE="${SCRIPT_DIR}/dos_smoke.bas"
EXPECTED="${SCRIPT_DIR}/expected/dos_smoke.expected"
DOSBOX_CONF="${SCRIPT_DIR}/dosbox-compat.conf"
if [ ! -f "$EXE" ]; then
echo "ERROR: $EXE not found. Run ./build_dos.sh 16 first." >&2
exit 1
fi
if ! flatpak list --app 2>/dev/null | grep -q com.dosbox_x.DOSBox-X; then
echo "ERROR: DOSBox-X flatpak not installed (com.dosbox_x.DOSBox-X)." >&2
exit 1
fi
# DOSBox-X (flatpak) can only access $HOME, not /tmp.
WORK=$(mktemp -d --tmpdir="$HOME" gw_dos_smoke.XXXXXX)
trap 'rm -rf "$WORK"' EXIT
cp "$EXE" "$WORK/GWBASIC.EXE"
cp "$SMOKE" "$WORK/SMOKE.BAS"
sed -i 's/$/\r/' "$WORK/SMOKE.BAS"
SDL_VIDEODRIVER=dummy SDL_AUDIODRIVER=dummy \
timeout 30 flatpak run com.dosbox_x.DOSBox-X \
-conf "$DOSBOX_CONF" \
-c "MOUNT C $WORK" \
-c "C:" \
-c "GWBASIC.EXE SMOKE.BAS" \
-c "EXIT" \
>/dev/null 2>&1
if [ ! -f "$WORK/OUT.TXT" ]; then
echo "FAIL: gwbasic16.exe produced no OUT.TXT under DOSBox-X" >&2
exit 1
fi
actual=$(mktemp)
normalized_expected=$(mktemp)
trap 'rm -rf "$WORK" "$actual" "$normalized_expected"' EXIT
sed 's/\r//g; s/[[:space:]]*$//' "$WORK/OUT.TXT" | sed '/^$/d' > "$actual"
sed 's/\r//g; s/[[:space:]]*$//' "$EXPECTED" | sed '/^$/d' > "$normalized_expected"
if diff -q "$normalized_expected" "$actual" >/dev/null; then
echo "PASS: gwbasic16.exe smoke test matches golden output"
exit 0
else
echo "FAIL: gwbasic16.exe output differs from expected"
diff -u "$normalized_expected" "$actual" | head -40
exit 1
fi