Closed mjburton11 closed 7 years ago
So immediate issues are that it can't find pandas. Maybe we don't need and the parameters could be returned as a dict.
And I think the max error of the fit should be returned with the constraint when doing a fit. But that's a design decision that needs to be made. Thoughts?
I'd stick the max and rms error into attributes of the returned ConstraintSet
; another factor pointing towards the convenience of a specialty FitConstraintSet
class.
I like the idea of the rms and max errors being attributes of the fit constraint set.
Am I seeing that the unit tests are only failing on Windows? @bqpd
Also, the pylint erros that I saw was that it doesn't like K
as an input...
@galbramc there's some DLL load errors on the windows side; might the libraries need to be updated?
@mjburton11 yeah, I'd just put a pylint-disable comment on that one.
@bqpd now the pylint error I'm getting is unable to import pandas. I thought we solved that by putting it inside the function?
yeah, but pylint tries to import it. You should be able to disable that error as well...
Test this please
Test this please
The DLL error was because the Jenkins script for gpfit was missing
REM turn on the anaconda console call activate.bat || exit /b 666
argh, merging this caused gpfit_Push to fail...
No, the gpfit push was missing
REM turn on the anaconda console
call activate.bat || exit /b 666
as well. I added it and it now passes.
@whoburg and @bqpd ready for review: return max error with rms error; allow for parameter conversion to pandas dataframe