is the fortran correct

[Alan J Hewitt (alanjhewitt)]

I am making a significant refactor of the radaer kernel, but psyclone will not parse it in my rose-stem tests.

I have had some success in identifying and fixing problems in my branch using fparser2 and fortitude but i still get the error
Parse Error: Failed to parse kernel code '/var/tmp/pbs.1951286.ehz100/radaer_api_la_side/run3/lfric_atm/cce_production-32bit/kernel/radaer_kernel_mod.F90'. Is the Fortran correct?

Are there any other tools that could help me find out what is wrong in my branch?

Also, just checking that we are using the latest version of fparser ?

My branch is here
https://github.com/alanjhewitt/lfric_apps/tree/radaer_api_la_side

The alternative is to add in small increments at a time, which is going to be a real pain as I am pretty much rewriting this module

Regards,
Alan

[Alan J Hewitt (alanjhewitt)]

OK, I spent some time with github copilot, and after a lot of back and forth it ran some tests on the latest version psyclone, and found the last build error. This was said by copilot

**Build-stage issue traced and fixed
I reproduced the LFRic/PSyclone failure and traced it to a parser-breaking typo in interfaces/physics_schemes_interface/source/kernel/radaer_kernel_mod.F90.

What was wrong
In the character-array declarations near the local name lists, the second declaration used acc_sol_volume_names where the code later expects cor_sol_volume_names. That breaks the old fparser frontend used by PSyclone during kernel parsing.**

It looks like psyclone will be updated at the met office soon after the next release.

I think I can close this.

Alan

[Steve Mullerworth (stevemullerworth)]

Note, while output from PSyclone is minimal, you can run fparser2 on the file, directly. Often that will output the parsed file up to the point of the error, which may allow you to to locate the problem.

[Alan J Hewitt (alanjhewitt)]

thanks Steve. I did run fparser2 directly on the file myself, but I didnt know that the currently installed version was an older revision that does not pick up all of the errors. I know that the new version is coming soon after the new lfric_apps release.

Its a really great tool, as is fortitude, and it is going to help everyone resolve these problems much quicker.