Hotfix: Update distance tables

Fixing the broken job introduced in !430 (merged). I'll put the tests back onto the schedule once it completes once.

added Bug label

This appears to not be working. I think I have a near-identical setup locally that isn't rewriting the files. Although, I did just spot an error with the phase marginalized version.

added 2 commits

• c6729d5f - Fix phase marginalization in lookups
• 27c20d0b - Add pip list

Compare with previous version

I'm looking at this and it looks like the difference is hardware dependent. If I run the test on the CIT cluster it rebuilds the files. There are a few (non-mutually-exclusive) solutions I can think of.

• As a short-term solution, I can try to push files built there as that is where most of the tests run and where we really need the shipped images to work.
• Longer-term, we can change the testing logic in Bilby to not match arrays, just to match the input prior distributions are the same.
• We could make it such that bilby_pipe never tries to overwrite the shipped lookup files. That might just hide inefficiencies down the road though.

@colm.talbot that is frustrating. Short term, I'd like to get the release done so people can use it on igwn. For that, I'd be okay to just relax the test for now or push the files that we know work?

Long term, maybe we shouldn't be shipping files if they are hardware dependent?

added 1 commit

• e3124e4e - build lookups on CIT

Compare with previous version

changed milestone to %1.0.6

mentioned in merge request !447 (merged)

mentioned in issue #238 (closed)

changed milestone to %1.0.7

Closing as this should be resolved by bilby!1150 (merged)

closed