When an ncml aggregation has some variables that don't depend on the aggregation dimension, those variables have origVar != null. However, the origVarshould not be used to read the data for that variable in this case, because the proxyReader is not the aggregation version (aggProxyReader). It has an iosp with a null raf, because this is based on looking up the urlPath (which does not have a dataset root so doesn't point to a file) instead of the using the aggregation location. I don't think this origVar should be used to read the variable data. This is not done in other services such as openDap.
I tested this with TDS, let me know if there is a good place to add a test in this repo.
This is a fix I made in the unidata fork of edal-java, so I thought it may be useful to port it here in case it is useful for this repo as well.
When an ncml aggregation has some variables that don't depend on the aggregation dimension, those variables have
origVar != null. However, theorigVarshould not be used to read the data for that variable in this case, because theproxyReaderis not the aggregation version (aggProxyReader). It has an iosp with a null raf, because this is based on looking up the urlPath (which does not have a dataset root so doesn't point to a file) instead of the using the aggregation location. I don't think this origVar should be used to read the variable data. This is not done in other services such as openDap.I tested this with TDS, let me know if there is a good place to add a test in this repo.