Handling of situations where all values for a variable are nans

Some summary IDS extraction was failing because some variables were filled with nans, they were completely masked and interpos was throwing an error since it was trying to interpolate empty arrays. When this happens, the variable is now filled with 1 and a warning is issued.

Note there is interpos_nan now that can be used, did you know?

No, I can change it and test it with that then. What would be the behavior with an array full of nans?

good question Should be the same as if providing a 0 length input to interpos Note that if less than 5 points or so, interpos uses linear interpolation (interpos has option to interpolate linear, quadratic or cubic spline, and for cubic spline with/without tension)

added 1 commit

• bb8f654a - Use interpos_nan to deal with nans

Compare with previous version

mentioned in merge request !189 (closed)

requested review from @lsimons

assigned to @mmarin

I've reviewed and tested, this is ready for merge. See RM ticket

approved this merge request

added 61 commits

• bb8f654a...569e08e4 - 59 commits from branch master
• 96c8799b - Handling of situations where all values for a variable are nans
• 99257b3a - Use interpos_nan to deal with nans

Compare with previous version

approved this merge request

merged