Error thrown during workflow _load_input_and_register method

[SWeav02]

I've been working with a student from Northwestern Polytechnical University in China to help them use the code for BadELF. He's run into an error that seems to be thrown during the base workflow classes _load_input_and_register method before any of the body of the code is run.

For some background on the BadELF workflow that's being used, it's made to run on DFT calculations that have already been completed. When running on DFT calculations that weren't completed using Simmate, and therefore not stored in the database, I found that you need to prepare the settings.yaml file slightly differently from normal. You place the settings file outside of the directory with the VASP output files. Then set that directories name as the directory tag and the relative path to the POSCAR as the structure tag. I'm planning to eventually make it so that users can use BadELF through the command line instead, similar to the Henkelman Bader code, but for now this is what's worked for me.

The student sent me the output log and settings file that they used, which I'm attaching here. I haven't run into this error before and I'm not sure how to fix it. I'm hoping there's a simple solution.

output.log
settings.txt

[jacksund]

I think your warren_lab apps works with v0.13.0 but not v0.14.0. So either pin your app to the earlier version, or update your code to work with v0.14.0.

I introduced a new rule in v0.14.0 from my changes in #466, which ended up as "bug". I put bug in quotes because this error would only happen if use_previous_directory=True is used incorrectly -- but I'll open an issue to improve the error message. The docs are here but could use some clean-up.

use_previous_directory is set to True for the population-analysis.bader.combine-chgcars workflow, and you're calling the workflow without any "primary input" or previous_directory. To fix this, change your code to something like this:

PopulationAnalysis__Bader__CombineChgcars.run(
    directory=new_directory,
    previous_directory=old_directory,
).result()

In your case, the old and new directories are one in the same. This doesn't follow "best practices" and might cause unexpected issues (because it'd try to copy files over itself). I'm planning on disallowing this eventually so beware.

# Not recommended and may causes issues
PopulationAnalysis__Bader__CombineChgcars.run(
    directory=old_directory,
    previous_directory=old_directory,
).result()

Update this everywhere you call PopulationAnalysis__Bader__CombineChgcars. Such as here