[Haskell-cafe] Trouble with asinh (c calls with Doubles) in Windows

[荒田 実樹]

The problem of asinh should be fixed in the C runtime library, so it
should be fixed in mingw, or by using another C library implementation
than mingw's.
Actually, Microsoft's recent C library (Universal CRT) seems to have
better math functions than mingw, so using it might be an option if
possible.
(cf. https://mail.haskell.org/pipermail/haskell-cafe/2021-April/133931.html
https://awson.github.io/ghc-nw/ )

As to why GCC has two different implementations of asinh -- the
implementation of math functions that GCC uses for constant folding is
MPFR, which is a bit heavy for linking to every program that it
compiles.

2021年9月7日(火) 18:01 David James <dj112358 at outlook.com>:
>
> Thanks again. I’ve added a note to the issue, and raised a bug against mingw. (And also updated another related one.)
>
>
>
> Is the right solution here to get it fixed in mingw? (And would that then be picked up in some future Haskell release?).
>
>
>
> I’m also still a bit confused about ming-w64 and GCC (which are all very new to me). Per Wikipedia “Mingw-w64 includes a port of the GNU Compiler Collection (GCC)”. So why does it have two different implementations of asinh? – one for use by GCC (that gives good results), and one that is called at runtime (that gives bad results)?.
>
>
>
> Thanks! David.
>
>
>
> From: arata, mizuki
> Sent: 04 September 2021 12:41
> To: David James
> Cc: haskell-cafe at haskell.org
> Subject: Re: [Haskell-cafe] Trouble with asinh (c calls with Doubles) in Windows
>
>
>
> FloatFnInverses is marked as ‘expect_broken’ on Windows:
>
> https://gitlab.haskell.org/ghc/ghc/-/blob/922c6bc8dd8d089cfe4b90ec2120cb48959ba2b5/testsuite/tests/numeric/should_run/all.T#L44-45
>
> And there’s a relevant issue: https://gitlab.haskell.org/ghc/ghc/-/issues/15670
>
>
>
> Mizuki
>
>
>
> 2021/09/04 18:46、David James <dj112358 at outlook.com>のメール:
>
>
>
> Hi - thank you for this. I was unaware of the “constant folding” in GCC (and I’m surprised it works for functions like asinh), but I can see that it explains the difference in behaviour.
>
>
>
> So I think this is a (possibly minor) bug that Haskell inherits from mingw-w64. I guess I should raise a GHC issue – though I’m not sure whether it would be best to try to fix within Haskell or within mingw-w64.
>
>
>
> Also, I think the FloatFnInverses.hs test doesn’t should be showing as a fail somewhere in the CI testing. (It doesn’t give the expected output when I run it on Windows). Do you know whether/where I can see that? (I don’t know what CI happens or how to view its output).
>
>
>
> Thanks again,
>
> David.
>
>
>
> From: arata, mizuki
> Sent: 03 September 2021 13:43
> To: David James
> Cc: haskell-cafe at haskell.org
> Subject: Re: [Haskell-cafe] Trouble with asinh (c calls with Doubles) in Windows
>
>
>
> Hi David,
>
> If I understand correctly, GHC uses mingw-w64’s libc implementation on Windows.
> Since mingw-w64’s math functions are not of very good quality, it is likely that asinh returns NaN for a very large input.
>
> As to why `asinh(1.7976931348623157e308)` in CAsinh.c produces (seemingly-correct) 710.4758, it is probably because the C compiler (GCC) uses a different implementation of asinh when doing constant folding.
> As a note, you may get a different (compile-time computed) result for `asinh(x)` if you set a more aggressive optimization flag.
>
> Mizuki
>
>
>
>