LonelyCat124

I changes lines 215-241 to: ``` members.append(node.children[dummy_args.index(ref.symbol)].member.copy()) childmember = node.children[dummy_args.index(ref.symbol)].member members[0].pop_all_children() while isinstance(childmember, StructureMember): childmember.detach() childmember = childmember.member members.append(childmember.copy()) members[-1].detach() childmember.detach() final_member = members[-1] indices = [] for index in...

My current issue isn't related to this (I think) so don't worry about that.

Ok, I've changed my mind and it definitely seems to be some weird behaviour between this and my other code. In `OMPTaskDirective`, I do `self._inline_kernels()`, which does: ``` kerns =...

I thought I'd checked that it hadn't been name changed, but I will check again tomorrow morning. I'll try to find a way to get the generated fortran code at...

I think we can do this check now using `_update_node`, and might be a good idea to do this earlier than later, and expand it to more types of nodes...

With all the changes to Otter I think we're going to try to swap to using PSyData to do the otter profiling. I'm going to leave this PR open for...

We're changing how this is done, closing the PR and deleting the branch as we will have a replacement.

This is going to attempt to use PsyData instead of separate transformations now we have full task support.

After discussing today, this is not going to use PSyData, but instead belong to the OMPTaskDirective/OMPTaskTransformation as an option. This will probably also have to belong to Single/Serial regions as...

I'm guessing the long-term plan for this is to have the call look up the routine symbol and check the intents (if declared) of the arguments to compute the accesses?...