Andy Porter

In fact, calling ACCRoutineTrans on the inlined kernel also fails because: ``` Transformation Error: routine 'matrix_vector_code_r_double' calls another routine 'MATMUL(matrix(:,:,ik), x_e)' which is not available on the accelerator device and...

I'm making good progress with the plumbing but am now hitting errors because of the plain `Symbol`s that are created in a kernel's symbol table for all of the reserved...

While working on this, I found this test (in lfric_kern_test.py): ``` def test_get_kernel_schedule_mixed_precision(): ''' Test that we can get the correct schedule for a mixed-precision kernel. ''' api_config = Config.get().api_conf(TEST_API)...

For the record, the code that we used to use to attempt to resolve the correct routine implementation was: ```python # The kernel name corresponds to an interface block. Find...

Thanks for thoroughly investigating Joerg. I like the idea of sticking to SemVer but I'm worried we're going to have e.g. 'alpha' versions of releases that will never actually exist....

Incidentally, when we build the fparser documentation it gives us:  which at least makes it very clear you've not got '0.1.5'.

> I am getting more and more the feeling that just adding `-dev` to master after a release would be easiest? I'm inclined to agree with you. KISS. No additional...

It's going to be 3.0. And yes, I agree with your suggestion. Probably it will be 3.0.1-dev and we'll change this to 3.1-dev as soon as anything significant gets merged....

I just hit this again in fparser - I was about to say I'd update the version number thanks to your pytest fix but then discovered that we already have:...

In fact, assignments involving character variables also cause problems. We could just exclude character literals from within OpenACC regions?