[Top][All Lists]

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [platform-testers] [Mpc-discuss] GNU MPC 1.2.0 release candidate

From: sisyphus
Subject: Re: [platform-testers] [Mpc-discuss] GNU MPC 1.2.0 release candidate
Date: Sun, 26 Jul 2020 22:49:12 +1000

Just to confirm that mpc-1.2.0-rc1 seems fine on native Windows 7, building in the MSYS2 (bash) shell.

I did strike something that confuses me wrt shared library (dll) builds.
Being confused by shared library builds on Windows is nothing new to me - and that's' the reason I shun dlls in preference to static libraries.

My mpfr-4.1.0 and gmp-6.2.0 dlls are in C:/msys/1.0/local_dyn/bin.
Therefore, 'make check' requires that C:/msys/1.0/local/bin be in my PATH environment variable - because  the test executables require that those 2 dlls be loaded.
Unfortunately, in the same folder there exists a libmpc-3.dll from a previous build of mpc-1.1.0.
Annoyingly, 'make check' insists on loading that pre-existing libmpc-3.dll instead of src/.libs/libmpc-3.dll -.even though.the (fully qualified) src/.libs directory occurs at the very beginning of the PATH.
With mpc-1.2.0 I'm finding that if I want 'make check' to succeed, then I first have to delete the pre-existing C:/msys/1.0/local_dyn/bin/libmpc-3.dll, otherwise tdiv.exe, tdot.exe & tpow.exe all fail, and ttan.exe hangs.

I thought I understood the method used by Windows 7 to determine which dll to load ... but apparently not.
It's probably a PEBCAK, but I cannot for the life of me work out why the mpc dll that occurs second in the PATH is being loaded instead of the mpc dll that occurs first.

Any thoughts on what I'm missing in trying to understand this weirdness ?

Configure command was:
$ ./configure  --disable-static --enable-shared LDFLAGS=-L/C/msys/1.0/local_dyn/lib CPPFLAGS=-IC:/msys/1.0/local_dyn/include --prefix=C:/msys/1.0/local_dyn


On Sat, Jul 25, 2020 at 12:02 AM Andreas Enge <> wrote:

after about two years, we would like to make a new release of GNU MPC.
The release candidate is available here:
   sha256sum 1f606dfa490c30beed05c47f595695c2d4205cc4f5c79cc9bdabd557f00f1b70

We would be grateful if you could test it and report potential problems to

A big change is that we rely on the recently released new MPFR version 4.1.0.

If all goes well, the release could become public at some time between
July 31st and the end of August.

In case you are curious, here are the news:
Changes in version 1.2.0:
  - Minimally required library version: MPFR 4.1.0
  - New functions: mpc_sum, mpc_dot
  - Several functions are more robust with a reduced exponent range
    (for example corresponding to IEEE 754 binary formats)
  - New tool mpcheck.

Thanks a lot,


Mpc-discuss mailing list

reply via email to

[Prev in Thread] Current Thread [Next in Thread]