[petsc-dev] Checking for valid C linker flags with Intel 19 on Cygwin

[Tim Steinhoff]

Satish, that should be exactly it. Thanks!

Am Mo., 1. Apr. 2019 um 16:04 Uhr schrieb Balay, Satish <balay at mcs.anl.gov>:
>
> I guess the following change will work with this compiler?
>
> diff --git a/config/BuildSystem/config/setCompilers.py b/config/BuildSystem/config/setCompilers.py
> index 3a616f0318..edd11894d7 100644
> --- a/config/BuildSystem/config/setCompilers.py
> +++ b/config/BuildSystem/config/setCompilers.py
> @@ -1004,7 +1004,7 @@ class Configure(config.base.Configure):
>          output.find('not recognized') >= 0 or output.find('not recognised') >= 0 or
>          output.find('unknown option') >= 0 or output.find('unknown flag') >= 0 or output.find('Unknown switch') >= 0 or
>          output.find('ignoring option') >= 0 or output.find('ignored') >= 0 or
> -        output.find('argument unused') >= 0 or
> +        output.find('argument unused') >= 0 or output.find('not supported') >= 0 or
>          # When checking for the existence of 'attribute'
>          output.find('is unsupported and will be skipped') >= 0 or
>          output.find('illegal option') >= 0 or output.find('Invalid option') >= 0 or
>
>
> Satish
>
> On Mon, 1 Apr 2019, Tim Steinhoff via petsc-dev wrote:
>
> > Hi,
> >
> > we moved from Intel 16 to Intel 19 compiler on Cygwin and it seems
> > that Intel has changed their error message, when PETSc is testing for
> > rpath flag.
> >
> > Intel 16 - PETSc rejects the flag correctly:
> > stdout: icl: command line warning #10157: ignoring option '/W';
> > argument is of wrong type
> >             Rejecting C linker flag -Wl,-rpath,/reps/petsc.git due to
> >
> > Now with Intel 19, PETSc thinks it is valid despite the warning message:
> > stdout: icl: command line warning #10148: option
> > '/Wl,-rpath,/home/jac/reps/petsc/petsc.git' not supported
> >             Valid C linker flag -Wl,-rpath,/home/jac/reps/petsc/petsc.git
> >
> > This leads to PETSc not recognizing that rpath is actually not
> > supported which will lead to problems later.
> >
> > Could you please add the new message to the check in
> > containsInvalidFlag (setCompilers.py) or should I create a PR?
> >
> > Thanks and kind regards,
> >
> > Volker
> >
>