ACCESS-ESM1.6: PR28-21 - code?

CC: @inh599
Prior to the hiccup we encountered with building from branches, our preferred/best configuration was PI-case2d.

We are hoping run this again with a runtime namelist tweak - so we can use the same executable (in PR28-21). However, re-visiting PR28-21 shows that it was built from a branch - circling back to the problem we later found.

So - where can I get the code that actually went into PR28-21 to verify that it is what we think it is OR is there another means by which I can have some confidence in what the PR28-21 model actually is?

Hi Jhan

I am not sure why pr28-21 doesn’t exist anymore in /g/data/prerelease/modules

See Spin up b · ACCESS-NRI/ACCESS-ESM1.6@f692a96 · GitHub - no new code was built for that action. Spack solved against a previously built version.

Did you want to know what version spack solved against ?

Tommy would be able to give you more details about PRs before 28-23 I think not being there. I’m not really concerned about anything other than what version of UM7 code was compiled BUT if PR28-21 doesnt even exist anymore there isnt much point. I’ll try to piece thing together to work out which code base to build a new PR from.

Do you know if PR28-21 might be spared from the whole complication of building from branches because it pre-dates modifications that @TommyGatti was making? And so the branch specified in the spec.yaml was indeed the branch it was getting code from. I

you know what - dont worry - Im just going to re-run a few cases sided by side and adopt a more robust naming convention for testing. Problem is we have code changes and config changes happening at the same time. The relevant working branch I have in ACCESS-ESM1.6 is Spinup_b. I’m assuming I can “git checkout -b” from here to branch from or do it from GitHub/issues end?

You can make new branches in the command line or through github, either way is good.

If its something to share with others, its best to make an issue first to describe what you are working on and use the issue number in the branch name

Btw - with a bit of spack magic we can figure this out if needed.