Paul Murphy

libdfp should follow the C TS-18661 and TR 24732. Which is just defining how the IEEE-754 decimal type works in C. More detail can be found in README.user. It's a...

I had looked briefly into this a few years back using "The Mathematical Function Computation Handbook" as a starting point. I can resume that work if you are interested. Properly...

Neat, thanks for volunteering. Admittedly, my modern C++ is a bit rusty, would this need to live inside libdfp? Though, my suspicion is the lack of intrinsic string functions in...

Would you have some representative test cases to demonstrate? I suspect this code was inherited and adopted from glibc. Looking at the test coverage, I am unsure if base 16...

Refreshing my knowledge of strtodN as standardized by TS-18661 part 2, only strings expressed in base-10 are acceptable and parsed. strto{f,d,ld,fN} (from C99 and TS-18661 part 3) convert base 10...

Issue #154 adds strfromdN

The plt bypass tricks for gcc intrinsics appear to no longer work when -flto is used. I am not familiar with how the linking machinery behaves when lto is enabled.

ppc64le has some interesting failures. `fabsd128` inlining into acosd128 is failing because the it preserves the asm usage of `fabsd128` but forgets it also needs to move the second half...

`__dpd_trunctdsd2` is a curious failure. It is a fairly small function ~7 instructions, and is used throughout libdfp. Based on what I can find, I don't think we can trivially...

I don't think there is a reasonable workaround to avoid local PLT calls when GCC intrinsic functions and `-ftlo` are present. The current mechanism is really an assembler level hack...