Reformat config machines

[jedwards4b]

Reformats config machines so that the primary file contains only the regex field for machine name matching. The
details for each machine are then read from files in machines/machinename/config_machines.xml
This is v3 of config_machines.xml, this change is backward compatible with v2.

Test suite: scripts_regression_tests.py
Test baseline:
Test namelist changes:
Test status: bit for bit

Fixes

User interface changes?:

Update gh-pages html (Y/N)?:

[jasonb5]

The config_machines_version3.xsd schema file is missing, otherwise this looks good.

[jedwards4b]

Added the config_machines_version3.xsd and the code to pass both schemas in and check the one based on the file version.

[billsacks]

Awesome. I like the new format of storing machine information! I have just one inline request for a bit more documentation on a function you changed. Some other comments / possible requests:

• There are a few changes in here unrelated to the main changes here (cprnc and some test stuff). Would it make sense to separate these into their own PR? I don't feel strongly about this.
• Does this lead to changes in the recommended (or possibly required) way that people store stuff in their personal .cime directory? If so, is there somewhere where this should be documented?

[rljacob]

Are these changes intended?

[jedwards4b]

These changes were due to a merge to master.

[billsacks]

Weird - usually a merge to master won't show up in PR diffs like this. Is the situation different for submodules like this?

[jedwards4b]

I agree this is a little strange and will make sure that the merge works correctly.

[billsacks]

Great, thanks!

[rljacob]

This is welcome since config_machines.xml can become a giant file. Does the code in the per-machine files look just like the XML blocks in the current combined file?

[jedwards4b]

@rljacob I made it backward compatible so no changes are required in e3sm. Yes the
format of individual files is nearly the same as the current block - the only difference is that the NODENAME_REGEX has been moved out of the block to the top level file.
Here is a link to the reformated cesm machines files.

[billsacks]

Thanks for the latest change!

[rljacob]

Is there a way to do one subdir with a uniquely named file per machine instead of a uniquely named subdir for each machine?

[jedwards4b]

What is your argument for that change? I considered it, but plan to treat config_batch.xml in the same way - so there will be a config_batch.xml file in each machine subdir.

[rljacob]

If there's going to be more then one file in each subdir, then it makes more sense. Does the subdir name have to match the string you pass to --machine ?

[jedwards4b]

Yes, currently the subdirectory name is expected to match the string in MACH=

[jedwards4b]

The cprnc submodule does seem to be out of sync with master, my attempts to fix the issue have failed - @jgfouca can you have a look? There should not be any changes in cprnc due to this PR.

[jgfouca]

@jedwards4b , I just merged #4545 , which should put us on the latest cprnc main commit. You should rebase or upstream this branch, take the cprnc commit that's on master, and the conflict will resolve.

[jedwards4b]

@jgfouca I just did a rebase again - it still doesn't seem to have worked.

[jgfouca]

It looks like the conflict is gone?

[jgfouca]

Oh, I see this PR is still trying to change the submodule. Try this:
git checkout origin/master $path_to_cprnc ; git add $path_to_cprnc; git commit. This assumes origin is the main repo, not your fork.

[jedwards4b]

Still no help:
git checkout origin/master CIME/non_py/cprnc ; git add CIME/non_py/cprnc; git commit
Updated 0 paths from b6c797123
Check Xml............................................(no files to check)Skipped
Fix End of Files.....................................(no files to check)Skipped
Trim Trailing Whitespace.............................(no files to check)Skipped
black................................................(no files to check)Skipped
pylint...............................................(no files to check)Skipped
On branch reformat_config_machines
Your branch is up to date with 'mydev/reformat_config_machines'.

Untracked files:
(use "git add ..." to include in what will be committed)
#DiagsCase.py#
#testcase.py#
CIME/#test_status.py#
CIME/.#test_status.py
CIME/ParamGen/requirements.txt
CIME/case/#check_input_data.py#
DiagsCase.py
ccs_config/
scripts/create_newcase.py.log
scripts/create_test.log
scripts/create_test.out
scripts/create_test2.out
scripts/test.out
testcase.py

nothing added to commit but untracked files present (use "git add" to track)
derecho5: ~/sandboxes/cime
:) git push mydev
Everything up-to-date

[jgfouca]

@jedwards4b , can you push this branch to the main repo? I'd like to check it out and modify it.

[jedwards4b]

I could but I shouldn't have to, you can clone or add my fork with
git remote add

[jgfouca]

I checked out your branch and somehow your cprnc submodule is still set to an older commit. This should fix it:

git fetch $esmci_remote
git cherry-pick $esmci_remote/jgfouca/jime_branch
git submodule update
git push

[jedwards4b]

That did it - thank you!