Two things to do here:
Check the tests that currently fail when make check is run. The handler should be producing correct output - check that and update the baselines if the handler is working correctly.
Look at test 405 which I’ve marked as expected fail because there is an issue with forming an aggregation on the outer dimensions of variables where the sizes don’t match.
More info about #2: I thought that using joinNew might be the problem - since the the two arrays were not the same exact shape ([x][y] versus [x][y]) maybe that was the reason for the failure. But these arrays work in a joinExisting aggregation just fine. However, when the Grids are renamed, there is a fail when reading data from the second of the two files. The problem is that the constraint using to access data from the second file uses size information gleaned from the first file. This is only the case when the Grids are renamed.
This is on the branch ncml_rename_test.
I think we might want to label this as a known bug, add a separate ticket for it, leave test 405 marked xfail and move on with merging this branch, since the issue reported regarding aggregations and renaming has been fixed for the kinds of data the reporters use, we should get that fix out to them.
These tests may have the correct values even though they 'fail': 87 93 97 101 105 109 113 117 405 406
Check the values using getdap and then update the baselines if the correct values are being returned.
I used 'besstandalone -c bes.conf -i testsuite.dir/406/test.bescmd | getdap -M -' to look at the values after first running 'make check' where 406 is the number of the test that failed.
The names to feed to getnerate_baselines if these tests are passing: