Hostlist

[hategan]

This makes (most) executors set a PSIJ_NODEFILE variable pointing to a nodefile, which is a solution for #200 and touches on ExaWorks/job-api-spec#148, although it would be good to figure out a way to formalize this in the spec.

It also contains a few other minor fixes.

[codecov]

Codecov Report

Merging #414 (37591e0) into main (d9d2370) will decrease coverage by 3.25%.
The diff coverage is 84.00%.

❗ Current head 37591e0 differs from pull request most recent head c952604. Consider uploading reports for the commit c952604 to get more accurate results

@@            Coverage Diff             @@
##             main     #414      +/-   ##
==========================================
- Coverage   73.65%   70.41%   -3.25%     
==========================================
  Files          79       78       -1     
  Lines        3575     3437     -138     
==========================================
- Hits         2633     2420     -213     
- Misses        942     1017      +75     

Files Changed	Coverage Δ
...c/psij/executors/batch/batch_scheduler_executor.py	75.49% <ø> (-1.41%)	⬇️
tests/_test_tools.py	70.00% <0.00%> (ø)
src/psij/executors/batch/slurm.py	40.98% <33.33%> (-14.19%)	⬇️
src/psij/launchers/script_based_launcher.py	91.11% <50.00%> (-1.12%)	⬇️
src/psij/executors/local.py	90.49% <84.09%> (-1.86%)	⬇️
tests/test_nodefile.py	95.45% <95.45%> (ø)
tests/plugins1/_batch_test/_batch_test.py	93.90% <100.00%> (+0.23%)	⬆️

... and 18 files with indirect coverage changes

[jameshcorbett]

Looks reasonable to me! The format of all the node files is the same--a hostname on each line or something?

[hategan]

Looks reasonable to me! The format of all the node files is the same--a hostname on each line or something?

In theory, yes.

[andre-merzky]

I like that one :-)