search

PolarGeospatialCenter / imagery_utils

Other
34 stars 10 forks source link

Change gdal version to 2.1.3 in qsub scripts to resolve numpy issue #28

Closed bakkerbakker closed 3 years ago

bakkerbakker commented 3 years ago

Testing the PR workflow with a quick fix to the GDAL version called in the qsub .sh scripts. The 2.1.1 version of the GDAL module did not have numpy installed, which caused the pgc_mosaic.py script to fail with the -pbs flag. Upon Erik's advice, I switched to 2.1.3, which resolved the issue.

ehusby commented 3 years ago

Looks good! When you make a PR that's worth running by other members of the team, typically just tag them under the "Reviewers" section instead of "Assignees" -- the latter should probably be used to appoint someone(s) in particular to make additional changes... (it makes more sense for Issues).

Thanks! Feel free to merge this when you're ready, and then pull changes into the common repo at V:\pgc\data\common\repos\imagery_utils. Typically you could do this either through Nunatak or Windows, but with Nunatak unavailable ATM, just a simple "Git Bash Here" in that folder through Windows followed by git pull should do the trick.

bakkerbakker commented 3 years ago

Got it! Somehow I missed Reviewers at the top but my eye landed on Assignees. I'll use Reviewers going forward. Thanks!

ehusby commented 3 years ago

Good point @clairecporter. As we run different scripts on the cluster, let's make an effort to test our pgc conda env on a case-by-case basis. Our pgc env uses GDAL version 3, and while hopefully there aren't changes between GDAL v2 and v3 that break scripts (besides the PROJ axis order thing), we should be careful to watch for any oddities in the results.

To test this in qsub scripts, we'd just replace the module load gdal/2.1.3 line with conda activate pgc.