Suppose that there are two files and they have the same variables. It's pretty easy to make an aggregation out of them. For example, see the first attachment. The first file is used as a template so we get variables from it and a variable for the aggregation. The problem is that the 'aggregation' covers several files but there's no way to know which file the 'non-aggregated' variables' values came from.
The fix is to remove the variables that are not part of the aggregation.
Should we put something in the code to enforce this.
Slav looked at this and we found that it's working fine in the current (1.15.2 - Jan 2019) version of Hyrax. The problem has likely been fixed in the HDF4 handler. When H4.EnableCF is set to true the DDS for one of the TRMM files is 'flat'
and has the lon and lat vectors and does not have the variables this big report references when it 'non-aggregated variables'. Compare the above DDS with the DDS returned when H4.EnableCF is false noting that three variables are there that are not present in the 'CF DDS' above:
So, this has been fixed.