Big endian serialization was broken because:
- it partially relied on `WORDS_ENDIAN` (unconditionally undef'd in cutils.h)
- endianness was not handled at all in the bc reader.
Modifications:
- remove `WORDS_ENDIAN`
- use `bc_put_u32()` / `bc_put_u64()` in `JS_WriteBigInt()`
- use `bc_get_u32()` / `bc_get_u64()` in `JS_ReadBigInt()`
- handle host endianness in `bc_get_u16()`, `bc_get_u32()`, `bc_get_u64()` and
`JS_ReadFunctionBytecode()`
- handle optional littleEndian argument as specified in
`js_dataview_getValue()` and `js_dataview_setValue()`
- split test_bigfloat.js from test_bignum.js
- make test_date() compatible with node
- document Date constructor string argument format:
should add test cases for invalid strings
- test_argument_scope(): only test this syntax error in strict mode:
`var f = function(a = eval("var arguments")) {};`
- Use uint32_t arithmetics and Standard conformant conversion to
avoid UB in js_math_imul.
- add builtin tests
- use specific object directories for SAN targets
ssize_t is not always available and the cast it was used in wasn't
necessary in the first place, the value already has the right type.
Note that the field malloc_limit is an int64_t in JSMemoryUsage whereas it
is a size_t in JSMallocState.
- add more cases of in place string concatenation
this temporary hack improves the microbench timing by 30% but
has little impact on the test262 timings.
- remove temporary fix for MemorySanitizer:
setting p->u.array.count to 0 silenced a warning in JS_GetPropertyValue on a
hacky test agains the length of fast_array object.
This hack was removed by commit c3635861f6.
- force evaluation order in `set_date_fields`
- fix evaluation error in test262/test/built-ins/Date/UTC/fp-evaluation-order.js:19:
unexpected error: Test262Error: precision in MakeDate Expected SameValue(«34448384», «34447360») to be true
MemorySanitizer complained about uninitialized reads in
the indexed property code path in JS_GetPropertyValue()
with JS_CLASS_MAPPED_ARGUMENTS objects.
