Sweep: last stale parity numbers (Getting-Started ~80%, Home 'approaching 100%') -> current 1614/1685, zero standard-C failures
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
63815b3
GCC Compatibility: refresh from v0.17-era 80% to the v0.38.0 reality
1327/1685 (~80%) -> 1614/1685 (95.8%; 99.4% of in-scope), with ZERO
standard-C failures — every classified class-(a) torture failure is fixed
(K&R defs, VLA parameters, native _Complex, struct-by-value ABI, bitfield
packing, computed goto, -0.0, ...). The 10 remaining failures are all GNU
extensions (SIMD >16B, aligned>16, empty-union ABI, __sync builtin), listed
in a table; the 61 formal skips explained with a pointer to the in-repo
classification + manifest. Approach section updated to the two-canon
methodology (gcc AND clang) plus the emitted-C oracle and the clang scope
filter. Stale known-gaps list (struct pass-by-value, ***p, __real__,
inline-asm-era items) deleted. Numbers from the 2026-07-23 container sweep
at the promoted v0.38.0.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
62d27be
Multi-file Projects: --project compile_commands.json is the primary flow
The page described only the old single-file #include-everything convention.
Now: the project driver (--project, implicit .json manifest, standard
compilation-database format from CMake/bear), -l linking, multi-TU native
executables (madc --project db.json -o app), with the SMAUG 51-TU example —
and #include composition demoted to the lightweight alternative it is.
Example verified against bin/madc v0.38.0 in all three lanes (JIT, implicit
json, native). Home doc-index line updated to match.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
dc4cf4c
Document the post-asmjit feature surface: C++ Support + Script Mode pages, auto-include, --emit=c11, expanded CLI flags
- NEW Cpp-Support page: classes/copy ctors, multiple+virtual inheritance
(Itanium ABI, RTTI, dynamic_cast), templates with real parse-once
instantiation, exceptions, spaceship operator, real libstdc++ interop —
with an honest not-yet list
- NEW Script-Mode page: main-less top-level statements, shebang, madc::sys,
exit-status semantics, native compilation of scripts (all examples
verified against bin/madc v0.38.0)
- Embedded-Headers: auto-header inclusion section (printf with no #include)
- Home: features list gains script mode, --emit=c11 standard-C output,
auto-include, direct ELF emission + -g; C++ bullet reflects the real
surface; doc index links the two new pages; Acknowledgments adds doctest
- CLI-Reference: flags table catches up with the real CLI (-c, -shared, -g,
--emit, --std, -I, --project)
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
800483f
Update to current reality: MIR backend, 'My Advanced Dialect of C', multi-arch framing
- Acronym: My jit-Assembled -> My Advanced Dialect of C (Home header, Getting-Started)
- Installation: asmjit v1.14 steps replaced by the madc MIR fork (MIR_COMMIT
pin, v<base>-madc.<version> release pairing, MIR_VERSION dependency file);
test-suite counts refreshed
- Home: features bullet + acknowledgments credit MIR/Vladimir Makarov (asmjit
kept as historical credit); x86-64 exclusivity softened everywhere — validated
on x86-64 Linux today, MIR backend also targets aarch64/ppc64le/s390x/riscv64
(ports, not rewrites)
- Native-Executables: direct ELF emission (no external toolchain), PIC + DWARF -g
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
75f04e2
Updated Namespaces (markdown)
1c3f436
Replace SMAUG-specific example with generic project in Multi-file Projects
2b8a522
Clarify that eval() compiles to native x86-64, not interpreted
20352ae
Add acknowledgment for asmjit and Petr Kobalicek
530bf20
Updated Why I created madc (markdown)
9bc8540
Link 'Why madc?' heading to backstory page
e8107ae
Updated Why I created madc (markdown)
a5f9b80
Updated Why I created madc (markdown)
487eff2
Link author name to GitHub profile
5ebfa6a
Add SMAUG authorship and MUD development background
eda5099
Polish backstory: add Delphi connection, SMAUG section, trim redundancy
6432698
Link 'Why I created madc' backstory from Not a toy section
b5092e8
Created Why I created madc (markdown)
e50a5b0
Fix eval() example: use int64_t overload instead of madc::value
2970356
Split embed/reuse into separate sections with distinct code examples
c61559e