Step: TopSpeed V3 corpus validation

Tag v3-topspeed-corpus. V3 suite 151/151, fixpoint OK (unchanged — this touches only the sidecar).

Why

After self-hosting, the next "normal" step was to exercise the front-end on a real corpus. The historical TopSpeed/JPI V3 corpus is the natural target: $HOME/bin/Dos/M2/TS-V3 — 662 files (352 .DEF + 310 .MOD).

V3's own M2.atg is the classic/Redux subset and would not take this dialect (pragmas (*#...), conditional compilation (*%...), H/C/ B literals, ::=, address constructors [seg:ofs]^, CLASS/IS, manifest constants, ...). The project already carries a sidecar grammar for it, compiler/src/TopSpeed-V3-M2.atg, whose deliverable is parsing the dialect (unlowered constructs end in the standard 230, parse-now / lower-later).

That sidecar had never been compiled and run over the full corpus.

What was done

Built the sidecar (tools/topspeed-grammar/build.sh) — copies the grammar + the V3 frames + SymTab/QbeGen/FileIO, runs CR -m -C, gm2 -fiso, links build/TSM2.
Fixed API drift vs the current SymTab/QbeGen (the sidecar predated several changes):
- SymTab.NewArray(e) → SymTab.NewOpenArray(e);
- SymTab.FixPending(t) / SetProcRes(t) now return BOOLEAN;
- QbeGen.EndModule now takes the module name.
Three real grammar gaps the corpus exposed:
- tagless variant CASE : T OF (anonymous selector) — TopSpeed accepted, the sidecar required a named tag;
- manifest constants without CONST (NAME = value; at declaration level);
- trailing | in a variant list (TRUE: …; | FALSE: …; |).
Corpus harness (tools/topspeed-grammar/corpus.sh) — runs TSM2 over every .DEF/.MOD in a scratch tree (so listings never touch the corpus) and classifies each file as parse-OK (only the deliberate 230/semantic markers) or syntax-bad (a real Coco/R error).

Result

	files
corpus	662 (352 `.DEF` + 310 `.MOD`)
parse-OK (230/semantic only)	631
syntax-bad	31

Coverage rose 590 → 627 → 631 as the fixes landed.

Remaining 31 syntax-bad

17 use (*%…*) conditional compilation — a lexer feature the sidecar does not implement; text inside a disabled block is parsed as code. (Pragmas (*#…) are already fine — they lex as ordinary (* … *) comments.)
14 assorted, e.g. a typed set constructor in an argument list (GetMenu("…", CharSet{'N','P','T'}, ch)), address constructors, and a few files with malformed/commented-out regions (END (*QMLB …). Full list in the harness output.

Reproduce

cd tools/topspeed-grammar
./build.sh          # -> build/TSM2
./corpus.sh         # -> 631 parse-OK / 31 syntax-bad

Notes

V3's own compiler (M2.atg) and the self-hosting path are untouched; the sidecar is a separate grammar built on demand.
The sidecar now tracks the V3 SymTab/QbeGen API; if those drift again, build.sh is where the small patch belongs.

Files

compiler/src/TopSpeed-V3-M2.atg (three grammar fixes), tools/topspeed-grammar/{build.sh,corpus.sh,README.md}, .gitignore (build dir), this doc.

summary_topspeed-corpus.md 3.2 KB تاريخچه خام