So it's failing in verify compiler config but it wasn't printing anything
about it until I went on a crash course learning as much about cmake as I
could in order to dump all the state variables. Once I started dumping the
state variables it started printing out the correct error at line 24.
I'll try deleting the cache there and see if that makes it recognized that
I actually do have current compilers installed now. I remain ideologically
opposed to brand new shiny in all the forms it can take but I will try to
integrate the energy drift fix if I can get past this.
On Tue, Mar 8, 2022, 06:02 David A Case <david.case.rutgers.edu> wrote:
> An update: we are (at least nominally) checking for a minimum version
> of gcc/gfortran in VerifyCompilerConfig.cmake:
>
> # check minimum GNU compiler versions:
> if (CMAKE_CXX_COMPILER_VERSION VERSION_LESS 6.0)
> message("-- Amber requires at least gcc-6.0")
> message(FATAL_ERROR "See https://ambermd.org/Installation.php
> for more info")
> endif()
> if (CMAKE_Fortran_COMPILER_VERSION VERSION_LESS 6.0)
> message("-- Amber requires at least gfortran-6.0")
> message(FATAL_ERROR "See https://ambermd.org/Installation.php
> for more info")
> endif()
>
>
> So the question is this: why is Scott seeing a "silent error" when he runs
> with g++/gfortran 5.4.0? Does he somehow have outdated code? (Scott: are
> you using Amber20?) Or is VerifyCompilerConfig.cmake not getting called,
> or
> not failing with a useful error message, as above?
>
> Some volunteer could check the latter possibility: try configuring on a
> machine with an ancient GNU compiler, and see what happens.
>
> ....dac
>
> _______________________________________________
> AMBER-Developers mailing list
> AMBER-Developers.ambermd.org
> http://lists.ambermd.org/mailman/listinfo/amber-developers
>
_______________________________________________
AMBER-Developers mailing list
AMBER-Developers.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber-developers
Received on Tue Mar 08 2022 - 07:00:02 PST
