Closed olejandro closed 8 months ago
I've udpated the benchmarks to cover a case like this:
PROCESS | NCAP_START | START | NCAP_START |
---|---|---|---|
PRC1 | 2012 | 2020 | |
PRC2 | 2012 | 2016 | |
PRC3 | 2016 | 2020 |
Currently the tests are failing on Demos7 due to additional records generated by an alias. The records appear in the correct order, so the input for GAMS will be correct.
Now that the additional records are gone the tests seem to be failing due to failing updated Demo 7 and Demo 7r benchmarks when run on main
. This makes sense because main
cannot handle the type of duplicate columns introduced in Demo 7. @siddharth-krishna I guess we'll have to override the check?
Thanks guys! Will revert the change to benchmarks version and add it together with a unit test in my next PR.
Implications of duplicate labels for operations in pandas are described here: https://pandas.pydata.org/docs/user_guide/duplicates.html#consequences-of-duplicate-labels
This PR ensures the data in a duplicated column is returned only once (and not e.g. twice if 2 columns with the same name are present) during processing.