The current implementation of the tests makes a binary comparison of a file generated with numpy and the file generated with xtensor-io. This comparison can fail wile the files are equivalent, meaning that each of this file can be read by both libraries and will give the same results.
This binary comparison depends on implementation details that do not change the data of the file, so maybe a more reliable test should be considered.
The tests of dump_npz have been commented out to avoid CI failure.
The current implementation of the tests makes a binary comparison of a file generated with numpy and the file generated with xtensor-io. This comparison can fail wile the files are equivalent, meaning that each of this file can be read by both libraries and will give the same results.
This binary comparison depends on implementation details that do not change the data of the file, so maybe a more reliable test should be considered.
The tests of
dump_npzhave been commented out to avoid CI failure.