Have added responses to inline comments now, and pushed related changes.

Some of your comments (those to do with adjoint_tests) are to do with changes that are not part of this PR - rather than looking at the "Files changed" section above, I suggest to look at the diff of this PR's branch with the previous PR's:

PLEASE NOTE - this is a follow-on to #132. The branch was created from #132's branch in my fork, but I can't make a PR into that branch because then I would be stuck in my fork. So, to look at the actual changes relevant to this PR alone, look at the diff of this branch with #132's branch: tom-j-h/lfric_apps@jelf_adjoint_test_tolerance_nml...tom-j-h:lfric_apps:align_jedi_lfric_tests_to_linear_model

Sorry - I did use your diff but managed to identify the wrong file when adding my comment, sorry for the confusion.

The reason there are some configuration options changed even though they don't actually have an effect is that previously there was a request to have the jedi_lfric_tests configurations match the linear_model/adjoint_tests ones exactly (except for JEDI-specific bits). If you'd rather those parts were unchanged then I'll do that instead. I see in the case of ls_option that it should definitely be different if it results in unnecessary code being run - thanks for pointing this out

The issue is the jedi specific bits are not obvious because we are not able to add that into the configuration. What we are doing is safe but it is a little confusing.

Similarly, the reason nwp_gal9 is done as an optional configuration which is included by default, is that this is how linear_model and adjoint_tests does it.

Thanks, yes Ive looked at the update. The nwp optional part is just pointing at newer files so I think the base configuration is effectively the nwp-gal9. Happy with what is done here now I've looked at the linear_model. I could have done that in the first place :-(.

As for the canned namelists, these were updated simply by copying over the generated namelists from rose-stem tests. I didn't spot those unnecessary differences you commented on - thanks.

Thanks, yes I assumed that was the case.

As for the other applications in jedi_lfric_tests other than jedi_tlm_tests and jedi_id_tlm_tests, I'm not really sure what these are for so left them alone. What do you suggest I do? I'm wary of it ending up being a lot of work with the deadline coming up.

Regarding the configurations:

rose-stem/app/jedi_lfric_tests test effectively runs the linear_model so should be based on the linear configurations. This should be a direct copy (for nwp_gal9 and runge-kutta) - the app is setup in the same way as the linear model (i.e. does not run via the emulator)

rose-stem/app/jedi_tlm_forecast_tl runs the linear_model via the emulator so should have similar changes to jedi_tlm_tests and jedi_id_tlm_tests

The other apps (jedi_forecast_pseudo and jedi_forecast) are based on the linear but they do not need to be kept in line with the linear.

It would be good to update jedi_tlm_forecast_tl and jedi_lfric_tests but its not essential since time is short. It can be done as a follow on.

I looked at the diff again tom-j-h/lfric_apps@jelf_adjoint_test_tolerance_nml...tom-j-h:lfric_apps:align_jedi_lfric_tests_to_linear_model, I noticed a few things that I mentioned in previous note that have not been updated. I think mainly due to confusion with me picking the wrong file. I add the notes here, can take a look?

1. I think the rose-app-default.conf can be removed?

2. In the configurations *.nml and *.conf, can you check:

i. File section looks like:

&files
ancil_directory='/data/users/lfricadmin/data/ancils/basic-gal/yak/C12',
checkpoint_stem_name='',
diag_stem_name='',
orography_mean_ancil_path='orography/gmted_ramp2/qrparm.orog',
start_dump_directory='',
start_dump_filename='',

ii. Initialization has:

init_option='analytic',
ls_option='analytic',

iii. Partitioning has (not sure how much it matters...):

&partitioning
panel_xproc=1,
panel_yproc=1,

Thanks Steve, everything addressed now, including updating configs for jedi_lfric_tests and jedi_tlm_forecast_tl. jedi_lfric_tests_developer group all okay except KGOs need updating: https://cylchub/services/cylc-review/taskjobs/tom.hill/?suite=align_jedi_lfric_tests_to_linear_model-jedi_lfric_tests_developer%2Frun2

Ready for rereview.

Thanks Steve, everything addressed now, including updating configs for jedi_lfric_tests and jedi_tlm_forecast_tl. jedi_lfric_tests_developer group all okay except KGOs need updating: https://cylchub/services/cylc-review/taskjobs/tom.hill/?suite=align_jedi_lfric_tests_to_linear_model-jedi_lfric_tests_developer%2Frun2

Ready for rereview.

Thanks for the updates. Happy with the changes but note that the KGO needs updating so some KGO checks are failing. This is due to an update to the configuration so it is expected and Im happy for the new kgo's to be added. Im not sure of the process related to that. Should that be done before sending to CR? @mike-hobson @tom-j-h

Thanks for bearing with me @ss421 ! I believe KGOs are updated by the Code Reviewer.

Looks like the ssd-bot has spotted that this can move to CR :-) now with @mike-hobson

This requires a small JEDI update. @tom-j-h will provide this. Captured in the next release PR here: https://github.com/JCSDA-internal/lfric-jedi/pull/1206

#132 is on which I think unblocks this one to be brought up to date and reviewed

#132 is on which I think unblocks this one to be brought up to date and reviewed

Thanks, now updated.