We just got our hands on Intel Fortran 14 here. (only a version behind now)
It looks like there's a null pointer dereference inspired seg-fault in
pFUnit when compiled with this version. Are you aware of anyone else using
this version of this compiler and observing this issue?
Ifort 14 is the main version I’ve been developing with for quite some time. Only last week did I upgrade to 15 on my laptop, but we still run 14 for the regression tests on the local cluster.
Please let me know exactly which version you are using and I’ll try to duplicate. There are very few pointers in the implementation, so it should not be too hard to track down. (Presuming it is not a compiler bug.)
[TLC]
Running the test through gdb and pulling a backtrace gives this:
It looks like a failure to call "tearDown" on the test which extends
ParameterizedTestCase. I do implement that method and it works on
GFortran. It's beginning to look like a compiler bug. [MH]
Last edit: Michael L. Rilee 2014-10-20
I have discovered some more about this. It looks like I can solve the problem by
explicitly exporting (i.e. specifying in a "public" clause) setUp and tearDown.
This doesn't make sense though as any methods not explicitly declared "private"
will be public. [MH]
Last edit: Michael L. Rilee 2014-10-20