...
- New containers (Docker, Singularity) available on ECR, Dockerhub, Sylabs
- Add missing dependencies for UK MetOffice
- Use generic target architecture to avoid "illegal instructions"
- spack-stack updates to compile fms, eckit, fckit, ecmwf-atlas on macOS with OpenMP
- Identified the bug in the IODA parallel HDF5 writer - need to come up with good strategy for replacing the colons, likely as part of the R2D2 mysql transition
- Jianjun Jin reporting he cannot read ioda_converter files with IDL and cannot dump them with ncdump (GSI to netCDF converters) - contact obs team
- R2D2 unit testing framework will be merged soon, good progress on R2D2 mysql database (but not yet ready to transition, will be announced later - full transition in next quarter for Skylab release v4)
- Dan H: ecbuild step takes a long time these days (number of repos? a lot of cmake files), on discover 10 minutes for fv3-bundle
- geos uses twice as many repos and cmake takes 40s
- Need to make sure it's not trying to download data as part of ecbuild step (done)
- Anna, David S, Steve V and Dom noticed this slowness as well
ALGO
- VADER code sprint report - attach slides from Steve Vslides:
View file name VaderSprintJEDISummary.pptx height 250 - Merged SABER testing framework refactoring
- Changes in yamls for bump, reference changes
- saber-data repository no longer needed, can be removed from bundles
- Anna to fill in PR in ??? that allows to do perturbations before/after qc, used in EDA
- Clementine debugging minimization in fv3-jedi using outer loops
- Francois/Dom are working on running UFS in core in ufs-jedi
- New application of JEDI for ensemble variance outputs ensemble variance and ensemble mean, but doesn't have the option of applying linear variable changes, please test and report, we want to keep this backward compatible
...