Test failure

Hello,

I'm trying install hdf5 on a HP Z800 (2x 6-core) with Scientific Linux
and Intel ifort, and I get 16 errors in the tests of "ph5diff". Can
anybody help?

The first error message is:

Testing ph5diff -v -d 5 -p 0.05 --use-system-epsilon h5diff_basic1.h5
*FAILED*
====Expected result (expect_sorted) differs from actual result
(actual_sorted)
    *** expect_sorted 2014-02-25 10:01:56.738317654 +0000
    --- actual_sorted 2014-02-25 10:01:56.737317608 +0000

···

***************
    *** 1,3 ****
    --- 1,10 ----
    +
    +

    +

    + = BAD TERMINATION OF ONE OF YOUR APPLICATION PROCESSES
    + = CLEANING UP REMAINING PROCESSES
    + = EXIT CODE: 1
      h5diff error: -d, -p and --use-system-epsilon options are
mutually-exclusive;
      Try '-h' or '--help' option for more information or see the h5diff
entry in the 'HDF5 Reference Manual'.
      use no more than one.
.
.
.
.
and there are 13 more of these, identical, except for the recorded time
of output. The message seems to be saying that the test was called
incorrectly. Could that have produced a bad result? Why didn't it fail
on bad syntax?

Any help would be much appreciated.

Ray

Ray,

The ph5diff tests compare the output from the execution with expected output.
The test you shown here is a verification of the tool detecting conflicting use of
parameters (can't use both -d and -p in the same run). The program will exit non-zero (1)
to signify failure detected.
Any chance that the mpiexec of the HP Z800 will output extra diagnosis messages
if it returns with non-zero exit code? If so, then the "errors" is not a real failure but
just a mismatch of expected output. We have no access to an HP system. If we had
access to an HP system, we could have taught the ph5diff tests to handle the "extra"
diagnosis messages.

Hope this helps.

-Albert
THG staff

···

On Feb 25, 2014, at 5:09 AM, Raymond D Wright <rdw@spacefleet.org.uk> wrote:

Hello,

I'm trying install hdf5 on a HP Z800 (2x 6-core) with Scientific Linux
and Intel ifort, and I get 16 errors in the tests of "ph5diff". Can
anybody help?

The first error message is:

Testing ph5diff -v -d 5 -p 0.05 --use-system-epsilon h5diff_basic1.h5
*FAILED*
====Expected result (expect_sorted) differs from actual result
(actual_sorted)
   *** expect_sorted 2014-02-25 10:01:56.738317654 +0000
   --- actual_sorted 2014-02-25 10:01:56.737317608 +0000
   ***************
   *** 1,3 ****
   --- 1,10 ----
   +
   +

   +

   + = BAD TERMINATION OF ONE OF YOUR APPLICATION PROCESSES
   + = CLEANING UP REMAINING PROCESSES
   + = EXIT CODE: 1
     h5diff error: -d, -p and --use-system-epsilon options are
mutually-exclusive;
     Try '-h' or '--help' option for more information or see the h5diff
entry in the 'HDF5 Reference Manual'.
     use no more than one.
.
.
.
.
and there are 13 more of these, identical, except for the recorded time
of output. The message seems to be saying that the test was called
incorrectly. Could that have produced a bad result? Why didn't it fail
on bad syntax?

Any help would be much appreciated.

Ray

_______________________________________________
Hdf-forum is for HDF software users discussion.
Hdf-forum@lists.hdfgroup.org
http://mail.lists.hdfgroup.org/mailman/listinfo/hdf-forum_lists.hdfgroup.org