Wuffs the Language has no floating point types, but some reference
implementations (notably the brotli encoder) make decisions with double and
float arithmetic and libm's log2. Porting them with byte-identical output
needs bit-exact floating point results.
This package implements f64 and f32 add, sub, mul, div, fma, comparisons,
min/max, integer conversions and f64<->f32 conversions on bit patterns,
with round-to-nearest-even, subnormals, infinities and NaNs, following the
structure of Berkeley SoftFloat 3. It also ports glibc 2.39's x86-64
__log2_fma log2 implementation (the variant selected on CPUs with FMA and
AVX2), following the shipped machine code's fused operations.
Verification: test/c/std/softfloat.c checks every operation against the C
compiler's native arithmetic on edge cases and random operands. The
script/manual-test-softfloat-*.c programs run 10^8 random operations per
operation group and compare log2_u64 against libm for every integer in
[1, 2^32) plus 10^7 larger values; all with zero mismatches.
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_011NoosQNe7DKWnqDCBeqw1p
The -O3 codegen is the same (https://godbolt.org/z/5ecWrEq77) but this
avoids compound literals, not part of C++11, and can trigger "compound
literals are a C99-specific feature [-Werror,-Wc99-extensions]".
Fixes#181
TestWriterChunked writes one file several times, splitting the contents
over Write calls differently each time, and checks the archive bytes come
out identical. The capped writer makes a runaway Write fail the test
instead of hanging it.
TestWriterDir writes a directory and then a file inside it, and reads
both back.
WriteHeader copied h into w.header before calling w.Write to emit the
GNUTYPE_LONGNAME block holding the name. Write only accepts TypeReg and
TypeGNUSparse, so a TypeDir header failed that call and left the Writer
permanently in an error state - every later entry on the same Writer
failed too.
Set w.header after the name block has gone out instead.
Write left w.bIndex at 512 once a full block had been written out, so the
next loop iteration took the "partial block" branch, copied zero bytes and
went round again forever, writing the same 512 bytes over and over. Calling
Write with sizes that aren't multiples of 512 is enough to hit this.
It also duplicated the last data block when a Write call happened to end
exactly on a block boundary, because flush then saw a non-zero bIndex and
wrote the block out a second time.