Fix prev_dir behavior in input set generator of MPGGAStaticMaker

[Andrew-S-Rosen]

Closes #995.

[janosh]

glad you caught this @Andrew-S-Rosen. we absolutely need a regression test. this is as a big a gotcha as they come.

maybe we want to pass inherit_incar=False here as well to be explicit. it already defaults to False but that's reliant on a default value two parent classes away in a different repo. better be safe and repeat

atomate2/src/atomate2/vasp/jobs/mp.py

Line 58 in 9600cef

default_factory=lambda: MPRelaxSet(force_gamma=True, auto_metal_kpoints=True)

[Andrew-S-Rosen]

@janosh: I was also wondering about that --- I had to dig through classes to see what value inherit_incar was being set to just to be sure. I will add inherit_incar=False to the other sets here that way when people make a new one, they will be more likely to ensure that this is still the case regardless of what happens upstream.

Definitely agree about the regression test. I wish I could be more help on this. If nobody is able to chip in, I'll see if I can carve out some time in the next week or two.

[Andrew-S-Rosen]

@janosh: Under-promise and over-delivering over here --- I added a regression test for the prev_dir behavior with MPGGAStaticMaker.

However, what is concerning is that there does not appear to be any pre-existing tests for MPGGAStaticMaker in test/vasp/jobs/test_mp.py so we may only be testing that Maker indirectly. That's for another PR though.

[janosh]

excellent, thanks a lot @Andrew-S-Rosen! 👍

[codecov]

Codecov Report

All modified and coverable lines are covered by tests ✅

Project coverage is 76.01%. Comparing base (96b2b82) to head (428ebd1).
Report is 2 commits behind head on main.

Additional details and impacted files

@@            Coverage Diff             @@
##            main     #996       +/-   ##
==========================================
+ Coverage   4.38%   76.01%   +71.62%     
==========================================
  Files        174      174               
  Lines      12671    12690       +19     
  Branches    1890     1892        +2     
==========================================
+ Hits         556     9646     +9090     
+ Misses     12081     2506     -9575     
- Partials      34      538      +504     

Files with missing lines	Coverage Δ
src/atomate2/vasp/jobs/mp.py	92.59% <ø> (+92.59%)	⬆️

... and 156 files with indirect coverage changes

[Andrew-S-Rosen]

Would it be possible to get a bugfix release after this PR? I want to reduce potential issues with users trying to reproduce MP settings and getting conflicting results, especially given how hard it is to track down...

[utf]

v0.0.17 should be released shortly @Andrew-S-Rosen. Thanks for reporting this.