1. Launch of tests associated with the code_aster solver#

It is necessary to have installed the salome_meca platform using the instructions available on the site `www.code-aster.org`_ < http://www.codeaster.org/> < http://www.codeaster.org/>``_after downloading the ad-hoc version.

The recipe procedure described below has been validated for versions of salome_meca starting with version 2021. The aim is to launch the embedded tests in the version (s) of the embedded c*ode_aster* solver (es).

Salome includes a script to launch tests. This script can be used to automatically launch the entire test base delivered and to carry out an assessment that is easy to compare to that shown in the quality sheet of the operating version (fascicle A0 of manual A of the code_aster documentation). Any deviations from the list of tests mentioned as ending in error must be reported to the R&D project team so that it can decide.

The command that allows you to start the recipe procedure is for example:

export ASTER_RESULTS_DIR =/directory/results/
mkdir -p $ ASTER_RESULTS_DIR
Salome test -L' ASTER_XX. *sequential.*verification'-j 8

The environment variable ASTER_RESULTS_DIR defines the location where you want to write the test case result files (allow about 3 GB of free space).

The number of tests launched simultaneously (8 in the example) must be adapted to the platform used and to the number of cores available, and ASTER_XX must be adapted to the version being tested (for example ASTER_15 .4.0 to isolate the tests from version 15.4).

Note:

If the tests are run on a remote machine, it is necessary to activate the « display access control », so you must use the -X or -Y option when connecting ssh to the machine under test: ssh -X machine_target

At the end of execution, the balance sheet is displayed in the terminal:

3360/3362 Test #7687: ASTER_14 .4.0_zzzz367a… Passed 17.67 sec

3361/3362 Test #7681: ASTER_14 .4.0_zzzz364a… Passed 68.67 sec

3362/3362 Test #7623: ASTER_14 .4.0_zzzz337b… Passed 146.09 sec

Aster 14.4 balance in salome_meca 2019.0.1

99% tests passed, 11 tests failed out of 3694

Label Time Summary:

ASTER_14 .4.0 = 116067.30 sec*proc (3694 tests)

ASTER_VERIFICATION_SEQUENTIAL = 116067.30 sec*proc (3694 tests)

SMECA_INTEGR = 201.06 sec*proc (8 tests)

Total test time (real) = 14623.85 sec

The following tests FAILED:

1372 - ASTER_14 .4.0_rtool01e (Failed)

1479 - ASTER_14 .4.0_sdll123d (Failed)

1730 - ASTER_14 .4.0_sdlx104a (Failed)

1846 - ASTER_14 .4.0_sdnv105c (Failed)

1851 - ASTER_14 .4.0_sdnv108a (Failed)

2696 - ASTER_14 .4.0_ssnl133e (Failed)

3569 - ASTER_14 .4.0_ssnv250e (Failed)

3599 - ASTER_14 .4.0_ssnv256a (Failed)

4231 - ASTER_14 .4.0_zzzz151a (Failed) ok, incorrect return code but no error

Errors while running CTest

You must then compare the result of the tests with the quality sheet of the associated version (cf. for example [A0.03.40] « Quality sheet of the operating version of code_aster: version 14 »).

For gap tests, it is necessary to run the test again using the following command (do not forget the environment variable ASTER_RESULTS_DIR):

Salome test -R ASTER_YY .y.y_xxxx -V

with YY.Y.Y the version of code_aster considered and xxxx the name of the test, for example:

Salome test -R ASTER_15 .4.0_zzzz151a -V

and to forward the log to the R&D project team so that it can decide on the severity of the discrepancy.