change type of results_root and inputs_root from args to string

[zacharyburnett]

resolves SCSB-114

this change could break downstream if people are assuming that results_root and inputs_root are passed as a list of strings

[zacharyburnett]

downstream usages:
spacetelescope/drizzlepac#1715
spacetelescope/jwst#8149
spacetelescope/romancal#1056

[pllim]

Why is this needed, especially if it is going to break existing usages?

[zacharyburnett]

Why is this needed, especially if it is going to break existing usages?

I think we can assume the inputs_root and results_root will always be a single path, and all existing usages of it are indexing the first entry unconditionally anyway.

[pllim]

Sure, but then all the downstream packages will need to do a version check of ci-watson and handle stuff accordingly. I just don't see the point. What was the motivation for this?

[zacharyburnett]

the JWST regression test jobs were failing at the results parsing stage, because they couldn't correctly parse inputs_root and results_root; however, it looks like this was solved by a PR to jenkins_shared_ci_utils:

• Fix quote replacement jenkins_shared_ci_utils#89

You're right, it might be more headache to fix this than it's worth. I'll close this