[petsc-dev] errors in master

[Barry Smith]

  Satish,

   Based on

Build on ps4 arch-mswin-intel-cxx-cmplx TESTMODE=testexamples DUMMYARGUMENT=foobar  Fri, Jul 21, 2017  2:20:41 PM 

******* Testing: testexamples_C *******
2,12c2,47
< *** PetscViewer[write] using MPI-IO ***
< *** PetscViewer[write] skipping header ***
< *** PetscViewer[load] skipping header ***
< *** PetscViewer[load] using MPI-IO ***
< VecCompare
<   min(a)   = +0.00e+00 [loc 0]
<   max(a)   = +2.81e+03 [loc 359]
<   min(b)   = +0.00e+00 [loc 0]
<   max(b)   = +2.81e+03 [loc 359]
<   min(a-b) < 1.0e-10
< Headerless read of data verified for: dmda-mpiio.pbvec

Build on ps4 arch-mswin-intel TESTMODE=testexamples DUMMYARGUMENT=foobar  Fri, Jul 21, 2017  4:21:04 PM 

# -------------
#   Summary    
# -------------
# FAILED 
# success 816/1237 tests (66.0%)
# failed 0/1237 tests (0.0%)
# todo 44/1237 tests (3.6%)
# skip 377/1237 tests (30.5%)
Fri, Jul 21, 2017  4:30:59 PM
******* Testing: testexamples_C *******
2,12c2,47
< *** PetscViewer[write] using MPI-IO ***
< *** PetscViewer[write] skipping header ***
< *** PetscViewer[load] skipping header ***
< *** PetscViewer[load] using MPI-IO ***
< VecCompare

   Am I suppose to conclude that MPI-IO doesn't work with Intel compilers and Microsoft implementation of MPI ?

   We have no configure tests that MPI-IO actually works; we merely test that certain MPI-IO functions link against the libraries.

   Please cook up a simple C usage of MPI-IO that crashes with Intel compilers and Microsoft MPI libraries. Open MPI IO file? Writing to it? Closing it? Opening it again to read? Whatever parts you need. Then you can have the MPI-IO test in mpi.py build and run the test before setting MPI-IO as available. Skip the run test if running in batch.

   Thanks

     Barry