Rupert Ford

In general there is a problem in distinguishing between arrays and functions in Fortran unless you are able to fully link the program and in general we are only parsing...

I've fixed the 0 or 1 instance problem but now break the addition of comments so I think I'll need to add a comment loop before incrementing the class pointer....

There appears to be a noticable improvement in performance for the example benchmark ... master: Time taken for parse = 36.27s Branch: Time taken for parse = 29.06s

When testing the latest version with LFRic algorithm code I am finding parse errors in code like ... ``` ... use x #if defined(UM_PHYSICS) use y ... ``` which is...

Interestingly comments between use statements seem to be OK but ifdef's are not. I thought we processed these in the same way. I guess there must be a difference somewhere.

OK, I now realise that comment, include and preprocess rules are added to the general blockbase rules, so if we do them in order they will not be picked up...

Comment restructuring seems to be working. All tests pass, LFRic code is being parsed OK with psyclone and psyclone tests pass after finding and fixing another bug in one of...

Added test, which seems to work as expected, giving the following (which preserves comment locations) ... ``` PROGRAM test INTEGER :: arg1 INTEGER :: iterator !comment_out 1 arg1 = 10...

I'm not sure why I didn't make this ready for review before so here we go ...

I started adding a test for strict_order and found that the implementation is not working correctly. For example ``` integer :: a use my_mod ``` should not match with Specification_Part...