[petsc-dev] broken nightlybuilds (next vs next-tmp)

[Jed Brown]

Satish Balay <balay at mcs.anl.gov> writes:

> On Sun, 5 Nov 2017, Satish Balay wrote:
>
>> > $ git fetch -p && comm -12 <(git branch -r --merged origin/next-tmp | sort) <(git branch -r --no-merged origin/master | sort)
>> >   origin/hongzh/copy_l2g_stencil
>> >   origin/jed/fix-dmcoarsenhookadd-identical
>> >   origin/knepley/feature-ex11-shock-test
>> origin/barry/remove-unneeded-use-of-petscdatatype
>
> Actually origin/hongzh/copy_l2g_stencil had issues [and so did
> origin/barry/remove-petscdatatype-dmgetworkarray] So the branches that
> got tested are:
>
> $ git fetch -p && comm -12 <(git branch -r --merged origin/next-tmp | sort) <(git branch -r --no-merged origin/master | sort)
>   origin/barry/delete-adjoint-trajectory
>   origin/barry/remove-unneeded-use-of-petscdatatype
>   origin/barry/use-mpi-datatype-F90Array
>   origin/jed/fix-dmcoarsenhookadd-identical
>   origin/knepley/feature-ex11-shock-test
>   origin/stefano_zampini/fix-missing-locks-tao
>   origin/tristan/mat-invert-block-diagonal-mat
>
> Jed,
>
> Is origin/jed/fix-dmcoarsenhookadd-identical destined for maint or master?

maint.

> Barry,
>
> There is one regression with: origin/barry/remove-unneeded-use-of-petscdatatype
> http://ftp.mcs.anl.gov/pub/petsc/nightlylogs/archive/2017/11/06/examples_full_next-tmp.log
> Build on grind arch-linux-opt-cxx-quad TESTMODE=testexamples DUMMYARGUMENT=foobar  Mon Nov  6 06:09:25 CST 2017 
>> [ 0]3200 bytes PetscViewerVTKFWrite() line 265 in /sandbox/petsc/petsc.clone-3/src/sys/classes/viewer/impls/vtk/vtkv.c
>
> Looks like dtype is getting reset before reuse.
>
> Is the following fix ok?
>
> Thanks,
> Satish
>
> ---------
> $ git diff |cat
> diff --git a/src/sys/classes/viewer/impls/vtk/vtkv.c b/src/sys/classes/viewer/impls/vtk/vtkv.c
> index 7f72665f9e..58caed1a62 100644
> --- a/src/sys/classes/viewer/impls/vtk/vtkv.c
> +++ b/src/sys/classes/viewer/impls/vtk/vtkv.c
> @@ -245,6 +245,7 @@ PetscErrorCode PetscViewerVTKFWrite(PetscViewer viewer,FILE *fp,const void *data
>  {
>    PetscErrorCode ierr;
>    PetscMPIInt    rank;
> +  MPI_Datatype   vdtype=dtype;
>  #if defined(PETSC_USE_REAL___FLOAT128)
>    double         *tmp;
>    PetscInt       i;
> @@ -265,10 +266,10 @@ PetscErrorCode PetscViewerVTKFWrite(PetscViewer viewer,FILE *fp,const void *data
>        ierr = PetscMalloc1(n,&tmp);CHKERRQ(ierr);
>        for (i=0; i<n; i++) tmp[i] = ttmp[i];
>        data  = (void*) tmp;
> -      dtype = MPI_DOUBLE;
> +      vdtype = MPI_DOUBLE;
>      }
>  #endif
> -    ierr  = MPI_Type_size(dtype,&dsize);
> +    ierr  = MPI_Type_size(vdtype,&dsize);
>      bytes = PetscVTKIntCast(dsize*n);
>  
>      count = fwrite(&bytes,sizeof(int),1,fp);