#51 Compilation error on tests with intel 13.1.0

[Albert Jornet]

I compile with intel 13.1.0 the current pFUnit release. I get the following error:

env vars:
export F90=ifort
export F90_VENDOR=Intel
export MPIF90=mpif90

Compilation:
make tests

Error message:
Test_XmlPrinter.F90(41): error #6404: This name does not have a type, and must have an explicit type. [RUN%TEST]
suite = newTestSuite('TestXmlPrinterSuite')
^
/tmp/ifortEXlYJ8.i90(179): catastrophic error: Internal Compiler Error: possible out of order or missing USE
compilation aborted for Test_XmlPrinter.F90 (code 1)
make[1]: [Test_XmlPrinter.o] Error 1
make[1]: Leaving directory `/home/orchidee04/ajornet/util/pFUnit/tests'
make: [all] Error 2

[Tom Clune]

We are unable to reproduce this error with 13.1.3, which is the patch we use for testing 13.1. I've asked my programmer to replicate with 13.1.0 and I'll let you know.

Is there any possibility you can use a later Intel compiler? I can attempt a patch for 13.1.0 but am about to leave on travel. May not be able to create something for a couple of weeks.

[Tom Clune]

Just after my post, I got the results for 13.1.0. It does indeed reproduce the issue. But I'll continue to hope that you can simply upgrade your compiler.

[Albert Jornet]

I am not sure they will upgrade it in a while. I moved to the 2.X for the moment. Thank you.

[Tom Clune]

OK. But there are enough "good things" in 3.x that I will attempt a patch after my travels.

[Albert Jornet]

This will be very cool. Thanks

[Tom Clune]

Sorry - that I've not moved on this as quickly as I previously indicated. The DC snowstorm a few weeks back messed up my travel schedule. I think I'll actually have some time to work on this this afternoon - but the network in my hotel might make a mockery of these plans.

[Tom Clune]

I just pushed a patch to the development branch that will allow pfunit to compile with intel 13.1.0. Note that it still crashes in its self tests for the RobustRunner functionality, but I suspect it will now work for you. Please let us know either way.

Note - there is some risk that this will cause problems in other compiler versions in which case I will revert the patch. But I'm reasonably optimistic and may not be able to check the results from the automated regression tests until tomorrow.

Cheers.

[Albert Jornet]

I moved to the development branch. It successfully compiles.

As you said, it fails during the tests. This is the exception:

forrtl: severe (174): SIGSEGV, segmentation fault occurred
Image PC Routine Line Source
tests.x 000000000190D8D5 Unknown Unknown Unknown
tests.x 00000000004C21E1 test_robustrunner 69 Test_RobustRunner.F90
tests.x 0000000000AB3182 testmethod_mod_mp 86 TestMethod.F90
tests.x 0000000000AB1E4A testcase_mod_mp_r 133 TestCase.F90
tests.x 0000000000AB200E testcase_mod_mp_r 143 TestCase.F90
tests.x 0000000000AC106E testresult_mod_mp 235 TestResult.F90
tests.x 0000000000AB1D3F testcase_mod_mp_r 120 TestCase.F90
tests.x 0000000000AD25BB testsuite_mod_mp_ 114 TestSuite.F90
tests.x 0000000000AD25BB testsuite_mod_mp_ 114 TestSuite.F90
tests.x 0000000000ACF580 testrunner_mod_mp 100 TestRunner.F90
tests.x 0000000000406ED1 main_IP_runtests_ 131 selfTests.F90
tests.x 0000000000403AA8 MAIN__ 13 selfTests.F90
tests.x 0000000000403A2C Unknown Unknown Unknown
libc.so.6 00000033D9A1ED5D Unknown Unknown Unknown
tests.x 0000000000403929 Unknown Unknown Unknown

Thanks!