f207d74aec
Four roadmap items: - codegen: fix parenthesized string comparison. emit_atom didn't consume the body of a string-literal token (`"`), so for PRINT (A$+B$ < "ZZZ") it emitted a 0 placeholder, advanced one byte, and left "ZZZ" to be reparsed as a variable + extra trailing tokens -- the binary then failed to link with `var_ZZ_sng` undeclared. emit_atom now skips to the closing quote. Separately, the left_type tracking in emit_num_prec dropped VT_STR after a string + string concat (becoming VT_SNG), so the string-comparison codepath skipped when the relational operator arrived. Preserve VT_STR through TOK_PLUS when both operands are strings. Verified: paren string-cmp now compiles and produces the same -1 / 0 result as the interpreter. - compiler: --no-gc-check and --fast-math optimization flags. --no-gc-check skips the per-line gwrt_check_line() (no string-pool GC, no Ctrl+Break trap). --fast-math drops the divide-by-zero guard on `/`; the divisor still goes through (double) so 10/0 produces inf rather than SIGFPE. Both threaded through codegen_opts_t and exposed in --help. --inline-arrays from the roadmap deferred -- larger refactor. - interp: raise static caps on 32-bit / Linux builds. vars 256 -> 1024, arrays 64 -> 256, MAX_FOR_DEPTH 16 -> 64, MAX_GOSUB_DEPTH 24 -> 128, MAX_WHILE_DEPTH 16 -> 64. Codegen FOR_STACK_MAX 16 -> 64. Analysis-pass caps: MAX_LINES 4096 -> 8192, MAX_VARS 256 -> 1024, MAX_GOTOS 256 -> 1024, MAX_DATA 1024 -> 4096, MAX_GOSUB_RET 256 -> 1024. 16-bit DOS keeps the original modest caps via #ifdef _M_I86 -- the MEDIUM model has a single 64KB DGROUP for all static data and the bumped sizes broke runtime startup under DOSBox-X. 16-bit binary grew from 128KB to 132KB from the offset_secs field plus DATE$/TIME$ shift code, well within the FreeDOS budget. - interp + codegen: DATE$ / TIME$ assignment via process-local clock offset. Was a no-op accept-and-ignore. Now sets gw.time_offset_secs (long), and DATE$ / TIME$ / TIMER readers apply it to time(NULL) before formatting. The OS clock is unaffected (would need root). Compiled-binary readers also reference gw.time_offset_secs since libgwrt shares the gw struct. Verified: PRINT DATE$; DATE$="12-31-1999"; PRINT DATE$ shows the expected before/after in both interpreter and AOT paths. After these changes: 72/72 interpreter tests, 68/68 compat, 63/63 compiler tests, DOS smoke under DOSBox-X all pass. Build clean on both Linux (cmake) and 16-bit DOS (build_dos.sh 16).
5.6 KiB
5.6 KiB
Roadmap
Completed
Ahead-of-Time Compiler (v0.16.0)
gwbasic-compile translates tokenized .bas programs to C source, then
invokes GCC to produce native executables linked against libgwrt.a.
Pipeline: .bas → gw_crunch() → analysis pass → C codegen → gcc → native binary.
63 of 63 eligible tests pass (100%) via tests/run_compiler_tests.sh.
The harness only skips hardware-dependent tests (graphics/sound/timer)
and CHAIN/RUN target files that aren't standalone. The compiler now
accepts unnumbered direct-mode programs by auto-numbering them.
Language coverage:
- All statements: PRINT, LET, IF/THEN/ELSE, GOTO, GOSUB/RETURN, FOR/NEXT, WHILE/WEND, ON GOTO/GOSUB, ON ERROR GOTO, RESUME/RESUME NEXT, DIM, DEF FN, SWAP, READ/DATA/RESTORE, INPUT/LINE INPUT, OPEN/CLOSE/PRINT#/INPUT#/WRITE#, FIELD/LSET/RSET/GET/PUT, BSAVE/BLOAD, SAVE/LOAD, CHAIN/COMMON, SCREEN, PSET/PRESET, COLOR/LOCATE/CLS, CIRCLE/DRAW/PAINT/PLAY, VIEW/WINDOW/PALETTE, POKE/OUT/WAIT, DEF SEG, RANDOMIZE, CLEAR, MID$ assignment, ERROR, KILL/NAME/FILES/SHELL/MKDIR/CHDIR/RMDIR, ENVIRON, LPRINT/LLIST, WIDTH, KEY
- All operators:
+-*/\MOD^ANDORXORNOTEQVIMP><=<=>=<>(including string comparison via strcmp) - All functions: math, string, file, conversion (CVI/CVS/CVD/MKI$/MKS$/MKD$), graphics (POINT/PMAP), system (FRE/ERR/ERL/TIMER/DATE$/TIME$/ENVIRON$/INKEY$)
- Token embedding for complex statements (PRINT USING, DEF FN, graphics, file I/O, MID$ assignment) with selective variable sync
- Division-by-zero detection, RNG matching (gw_rnd), ON ERROR GOTO via setjmp/longjmp
Optimizations:
- Constant folding (compile-time arithmetic on literals)
- Dead code elimination (skip statements after GOTO/END/STOP)
- FOR step=1 elision (var++ instead of step variable, simple comparison)
- Fast-path expression emitter (skip buffering for common case)
- Selective variable sync in delegated statements
- REM-line skip (no runtime check for comment-only lines)
Hardware I/O Simulator (v0.15.0)
Implemented in portio.c / portio.h following the virmem.c dispatch
pattern. Emulates 8253 PIT channel 2 (speaker frequency), PPI port B
(speaker on/off with continuous tone via PulseAudio), CGA mode/color
registers, game port (joystick stub), and COM1 serial (transmitter-ready
stub). Default: reads return 0xFF (floating bus), writes discarded.
Also in v0.15.0: 100% token coverage (all 144 GW-BASIC tokens handled), string space pool with compacting garbage collector, RESET, ENVIRON/ENVIRON$, ERDEV/ERDEV$, IOCTL/IOCTL$, LCOPY, DATE$/TIME$ assignment, CALL, COM.
Jupyter Kernel (v0.15.0)
gwbasickernel/ -- Jupyter notebook kernel using the persistent subprocess
model with sentinel protocol.
- Inline Sixel graphics -- pure-Python Sixel decoder renders SCREEN commands as inline PNG images in the notebook
- INPUT statement support via Jupyter stdin protocol
- Pygments syntax highlighting for code cells
- Tab completion for all GW-BASIC keywords
- Magic commands:
%reset,%timeout,%new
Install: pip install -e . && gwbasickernel-install --user
Compiler Memory Safety (v0.17.0)
--warn, --safe, and --safe=sanitize flags for the ahead-of-time compiler.
--warn-- static analysis: uninitialized variables, GOTO to nonexistent line, unreachable code detection. Zero runtime cost.--safe(implies--warn) -- checked integer arithmetic viagw_int_add/sub/mul/neg(raises Overflow instead of wrapping), enhanced array bounds diagnostics with variable names and line numbers, GOSUB stack overflow diagnostics, ABS/SGN type-preserving codegen, string pool GC pinning infrastructure--safe=sanitize-- above plus-fsanitize=address,undefinedpassed to gcc
DOS / FreeDOS Target (v0.17.0)
Cross-compiles to DOS using OpenWatcom V2. Two targets:
- 16-bit real-mode (
Makefile.dos16): 128KB standalone MZ executable, MEDIUM memory model, far-heap TUI screen buffer, no DOS extender required - 32-bit DOS/4GW (
Makefile.dos): 175KB LE executable, flat memory model, requires DOS4GW.EXE extender
Tested on FreeDOS 1.4 via QEMU.
Next Up
Compiler Optimization Flags
--no-gc-check-- skipgwrt_check_line()per-line calls (no string pool GC, no Ctrl+Break check) for maximum throughput--inline-arrays-- emit direct array indexing for statically-DIMmed arrays instead of runtimegwrt_array_elem()lookup--fast-math-- skip division-by-zero checks, allow unsafe float ops-O0through-O3-- compiler-level optimization tiers mapping to different sets of codegen optimizations (constant folding, dead code elimination, FOR step=1 elision, fast-path expressions)
IDE Integration
- VS Code extension -- syntax highlighting (TextMate grammar), snippets, run/debug tasks, integrated terminal runner
- JetBrains plugin (IntelliJ/CLion) -- syntax highlighting, code completion,
run configurations, debugger integration (breakpoints via
STOP, variable inspection), structure view (line number outline)
Known Limitations
- Static caps -- 32-bit / Linux builds: 1024 variables, 256 arrays, 64 FOR nesting, 128 GOSUB nesting, 64 WHILE nesting. 16-bit real-mode DOS keeps the original modest caps (256 / 64 / 16 / 24 / 16) because the MEDIUM model has a single 64KB DGROUP for all static data.
CALL/CALLS(machine code execution) raises Illegal function callDATE$/TIME$assignment shifts the program's view of the clock via a process-local offset; the OS time is unaffected (setting the OS clock would require root)- Device stubs (
ERDEV,IOCTL,COM,LCOPY) return defaults