New ModelArrayRef

[timspainNERSC]

Will close #249

[TomMelt]

ICCS won't be able to meaningfully review this PR but the plan is to go ahead with the merge and rely on comprehensive integration tests (part of Milestone 2) to verify the correctness of this branch. After this branch is merged into develop we want the remaining features e.g., MPI and XIOS to be based off of the develop branch.

Going forward the aim is to keep features branches small and self-contained so that they can be merged back into develop in a timely fashion.

[TomMelt]

approved on basis of my previous comment. The code has not been reviewed but we will use the next milestone (adding integration tests) to verify the "correctness" of this branch.

[MarionBWeinzierl]

Just to flag: When running run_simple_example, this branch gives a

double free or corruption (out) 
Aborted (core dumped) 

error. Similarly, after the rebase in #331, that branch gives the same error, while commit a5e4b77 is fine. Trying the develop branch with the same test script gives a Segmentation fault (core dumped)

[TomMelt]

@timspainNERSC when you are back, can we fix the simple example scripts e.g. run_simple_example.sh. They do not run on this branch.

[timspainNERSC]

@timspainNERSC when you are back, can we fix the simple example scripts e.g. run_simple_example.sh. They do not run on this branch.

Interestingly, run_simple_example.sh does run on my computer, but only produces 1x30 output. Which it really shouldn't.

In what way do they not run? Segmentation fault? NetCDF error?

[TomMelt]

@timspainNERSC I recommend compiling with these flags -g -fsanitize=address -fno-omit-frame-pointer to see if there are any memory errors. This should help find them.

Also you can try running the code in gdb. when it crashes you can type command bt to get a backtrace. Happy to meet this week to discuss in more detail.

[MarionBWeinzierl]

With #425 being merged, do I understand it correctly that now #383 needs to be merged, and then this PR will be merged and upcoming problems solved as they come up? Or are we waiting for another issue/PR?

[timspainNERSC]

With #425 being merged, do I understand it correctly that now #383 needs to be merged, and then this PR will be merged and upcoming problems solved as they come up? Or are we waiting for another issue/PR?

I have just been through the priority PRs and marked their dependencies. As of today, this PRs #250 (this), #383 and #394 are able to be reviewed and merged. This is the most important as it impacts so much of the rest of the code.