Currently, when you call get_data() on an AggregatedPieces, the argument order is aggregated_experiments then independent_analyzers. The problem is that this is the opposite order that the analyzers will be run. The reason it's like this is that users may not want to run any independent analyzers, in which case they would simply provide the first argument and use the data kwarg to supply input data. Now, I think that situation will be sufficiently rare that it will cause less grief to supply an empty list as the first argument.
Since this requires breaking the API, I need to reverse these, putting them into order-of-execution, before VIS 2.
Currently, when you call get_data() on an AggregatedPieces, the argument order is
aggregated_experiments
thenindependent_analyzers
. The problem is that this is the opposite order that the analyzers will be run. The reason it's like this is that users may not want to run any independent analyzers, in which case they would simply provide the first argument and use thedata
kwarg to supply input data. Now, I think that situation will be sufficiently rare that it will cause less grief to supply an empty list as the first argument.Since this requires breaking the API, I need to reverse these, putting them into order-of-execution, before VIS 2.