Sergi Siso

> One thing that doesn't quite seem to fit compared with other transformations is the _directive class member. Since this can make many directives am I ok to just not...

Yes, I would go for the region that we fully understand (no Codeblocks). If we then find a use case with codeblocks we have two options: generalise the transformation if...

To clarify: - Nemo: inlined kernels where you can rely on dependency analysis/explore the tree to know what is safe to parallelise. - Gocean (NemoLite2D): coded_kernels but is direct addressing...

I have the same snippet of code that you added in the transformation apply in my NEMO script. So this will also work for me to simplify the script. I...

This is ready for review. Probably one for @arporter

@hiker I brought it to master and fixed some additional pylint and pycodestyle issues. This is ready for next review.

I started #1844 to fix these issues. I could deal with the duplicated declarations when the same kernel is in multiple invokes in the same file and a more subtle...

>I think we should allow/make an exception for kind parameters. In https://github.com/stfc/PSyclone/issues/924 I copy any imports over to the call site, even if they are in module scope. Ok, I...

I vote for psyclone without "-api XXX" to default to generic code. I am undecided yet about what I prefer for the config file. I am happy to take the...

As discussed, I now think the transformation was trying to do two things at the same time and the l/unbound->domain parameter can be a completely separate transformation that can also...