No test cases have <access type="discover"> blocks

[atz]

This would be considered an error in production. We should decide how we want to behave if given such data and most test cases should be updated to include an <access type="discover"> block.

[jmartin-sul]

FWIW: i see lots of <access type="discover">-related testing in spec/index_elements_spec.rb.

but that's just testing the index_elements stuff. what other behavior should be tested that's not getting tested at the moment?

[LynnMcRae]

Per conversation with John, the indexing logic would show the inconsistency (the error I think Joe refers to) by showing the object was declared both dark (per access=discover) as well as world/stanford/etc per access=read. This might initially alarm a collection owner who thinks their collection is all dark, but good that it would expose the contradictory data. We don't have any exposure issue here since the access=discover block determines if the PURL exists at all; any expression of object access after that would be meaningless as there would be nothing there to apply it to.