BFM-decomp/docs/readability.md

Readability of the matched C — the Gen3 series

Generated by tools/readability_progress.py --snapshot; never typed (R75). Companion to docs/levers.md, which counts the compiler-forcing levers. This one counts the two things that are wrong with the source even where no lever remains: call declarations that are FALSE, and memory read through raw pointer casts where the original had a struct. From Phase 37 the row carries the type census's columns (tools/type_census.py): all four dereference forms, the typed cast-member intermediate, the definitions and the map.

date	label	commit	lying declarations	of which ()	of which (void)	in a pinned body	raw cast derefs (one form, raw text)	struct member reads	raw derefs, four forms	typed cast-member	struct defs (canonical / in .c)	layouts	duplicate classes / names	map types (% sites) K&R sites (marked, apart)	cross-binary (apart)
2026-09-10	P36 S102 — after the 16,759 free declaration repairs	51028a78d	94001	86701	7300	461	414148	173286	—	—	—	—	—	—	—	—
2026-09-11	P36 close	7a7f78e25	98648	91357	7291	33	411850	177902	—	—	—	—	—	—	—	—
2026-09-11	P37 T1 — the type census (four forms, definitions, the map)	23fc5d215	98648	91357	7291	33	411850	177902	503016	36681	7261 (1179 / 5995)	525	206 / 2789	18760 (99.2 %)	—	—
2026-09-12	P37 T2 — the census after the audit's parser fixes (attributes, dead guards)	da26c05d6	98648	91357	7291	33	411850	177902	503016	36681	7255 (1179 / 5989)	527	206 / 2823	18760 (99.2 %)	—	—
2026-09-12	t4	274223396	8830	2997	5833	13	411848	177902	503013	36681	7255 (1179 / 5989)	527	206 / 2823	18741 (99.2 %)	5002	—
2026-09-12	t4-close	bd7d32ec2	1796	1373	423	1	411820	177902	502985	36681	7255 (1179 / 5989)	527	206 / 2823	18213 (99.2 %)	5262	418
2026-09-30	P38 close	c82a24899	229	216	13	0	411816	176269	502983	36853	1650 (1021 / 600)	523	53 / 496	19122 (99.2 %)	5293	418
2026-10-02	P3.39 T5: forms I/M closed	240775039	229	216	13	0	260306	320402	336219	173103	1650 (1021 / 600)	523	53 / 496	17965 (98.8 %)	5293	418
2026-10-02	P3.39 T11: chart catch-up	e7203a358	229	216	13	0	260306	320402	336219	173103	1650 (1021 / 600)	523	53 / 496	17965 (98.8 %)	5293	418
2026-10-02	P3.39 T6: lying declarations ledgered (229 -> 11 ledgered)	9aa588736	11	2	9	0	260306	320402	336219	173103	1650 (1021 / 600)	523	53 / 496	17830 (98.8 %)	5293	418
2026-10-02	P3.39 T7.1: GTE to one spelling (direct GTE 492 -> 0, per-TU gte defs 150 -> 0)	72e4c48c8	11	2	9	0	260306	320402	336219	173103	1650 (1021 / 600)	523	53 / 496	17878 (98.8 %)	5293	418
2026-10-02	P3.39 T8: lever residue TSV (2433 rows, reading A)	21f3aee1b	11	2	9	0	260312	320402	336225	173103	1650 (1021 / 600)	523	53 / 496	17677 (98.8 %)	5293	418
2026-10-02	P3.39 close: cast campaign, reading A	3a9377c92	11	2	9	0	260312	320402	336225	173103	1650 (1021 / 600)	523	53 / 496	17968 (98.8 %)	5293	418

How to read it. A lying declaration is a call site whose in-scope declaration names fewer parameters than the callee's own definition. It is not a style problem: on this processor an argument travels in a register, so a dropped argument removes an instruction, and Phase 36 found register pins inserted to fake exactly that instruction. Raw cast derefs is the struct debt — a struct cannot be recovered from the binary, only inferred from consistent base+offset+width evidence, so this figure falls as that work lands.