Closed TingLei-NOAA closed 1 month ago
It looks like a bunch of things were just copied from GDASApp to here without 1) thinking if they were relevant to the regional system or 2) renaming variables, scripts, etc.
I think this needs to be cleaned up before it gets merged or else these things will persist.
Yes, in this rdas re-organizing, I preferred keeping/copying components in gdasapp into rdas if they didn't cause much troubles for radas building. I (or other RDAS developers) would clean them through this PR process and future use/developments of radas.
@TingLei-NOAA I think we (@ShunLiu-NOAA @danholdaway ) should chat soon. GDASApp is containing the GDAS-specific implementation of JEDI, and will be where releases are tagged, and the source code is controlled with hashes/tags. I am nervous about blindly copying things for RDAS, as these two systems will be on a different upgrade cycle, and thus need different modulefiles, different hashes, and because of MPAS vs FV3 and other reasons, different utilities (like blending would not need to be in the global)
Ting, Thank you for working on this. @CoryMartin-NOAA @danholdaway, I have the same concerns about how to keep RDASApp and GDASApp on the same page.
@TingLei-NOAA I think we (@ShunLiu-NOAA @danholdaway ) should chat soon. GDASApp is containing the GDAS-specific implementation of JEDI, and will be where releases are tagged, and the source code is controlled with hashes/tags. I am nervous about blindly copying things for RDAS, as these two systems will be on a different upgrade cycle, and thus need different modulefiles, different hashes, and because of MPAS vs FV3 and other reasons, different utilities (like blending would not need to be in the global)
Sure. a clear plan for how to use rdas is needed. The current PR is supposed to make rdas could at least work (be able to be built on hera/orion) by, at least, EMC developers for being now. To use the current heads of submodules rather than using hashed commits in gdasapp (or rdas), a submodule updating script is added : checkout_submodules_branches.sh.
@TingLei-NOAA I think we (@ShunLiu-NOAA @danholdaway ) should chat soon. GDASApp is containing the GDAS-specific implementation of JEDI, and will be where releases are tagged, and the source code is controlled with hashes/tags. I am nervous about blindly copying things for RDAS, as these two systems will be on a different upgrade cycle, and thus need different modulefiles, different hashes, and because of MPAS vs FV3 and other reasons, different utilities (like blending would not need to be in the global)
FWIW, I am with @CoryMartin-NOAA here. The RDASApp should use GDASApp as a guide and add features that are needed and build on them, rather than import everything and then spend time removing them later.
@TingLei-NOAA Can we close this PR and open an new PR to merge other changes that you want to do to the current "develop"
This PR is closed for https://github.com/NOAA-EMC/RDASApp/pull/11 has been merged into develop branch. Thanks to all who had contributed to the discussions on this PR.
Following the current GDASapp which had switched to Git submodules to manage components, re-organize RDAS repo including updating its' module setup according to the current GDASapp. The mpas-jedi related components are also added while they are not included in the current GDASapp. The building of this RDAS bundle had been successfully done on Hera . More tests are to be done including a complete/clean "clone" and "build" process and on orion. But this branch could be used as an starting point for current work on RDAS.