Closed arunkannawadi closed 2 years ago
Thanks Arun. This seems sensible to me. One thing I'd like to see is a new unit test that fails with the old version of the code, but passes now with your changes.
Also, it looks like lots of the current unit tests are failing, so there is some backwards incompatibility in this for you to figure out.
Hoo, that's a lot of failures! I'll look into the errors and write aa new unit test as well.
I didn't add a new test case per se, but modified the existing ones so that they'd fail without the code changes in the PR. Tests passed in my fork: https://github.com/arunkannawadi/TreeCorr/runs/6435598018?check_suite_focus=true
I think there 1 or 2 lines that are not covered by the tests. I didn't run it locally, so I'll wait for codecov to pick it up when you run the tests here and update the tests to maintain the 100% coverage.
Hi Mike, the main intention of this PR is to be able to generate a
BinnedCorr2
object by passing all four ofnbins
,min_sep
,max_sep
,bin_size
but with exactly one of them setNone
. I've made the changes forBinnedCorr3
as well and modified some helper functions to not check if akey
exists in a config/kwarg dict but instead check if it has been explicitly set. I'm assuming that nowhere intreecorr
None
has a special meaning.