This is the mail archive of the mailing list for the GCC project.

Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]
Other format: [Raw text]

Re: ERFC_SCALED compile-time simplification

I hope this is temporary and you will be working on getting this into MPFR
and will remove this once GCC requires a new enough MPFR; MPFR has much
more expertise on reviewing implementations of such functions and we'd
generally prefer to rely on their expertise rather than maintaining such
implementations locally.

That would be the better course, of course. I'd welcome such inclusion into MPFR, would adapt gfortran accordingly (which would be trivial); in the meantime, I'll file a feature request for MPFR.

Now, regarding me working to get this into MPFR, the thing is getting it into gfortran and into MPFR have different requirements: MPFR requires a proven, works-for-all-sizes algorithm, including correct (and proven) bounds for the various domains, while for gfortran, all I need is something that works for precisions implemented on GCC- supported targets. While I am competent enough for the second one, the scale of the first is outside my capabilities.


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]