Guidelines for LIDVIDs for regression tests

[wkiri]

Inspired by our review of the namespaces, I suggest that we identify additions to the guidelines for creating test cases that are provided here:
https://pds-data-dictionaries.github.io/development/ldd-test.html

1. I recommend including guidelines for specifying LIDVIDs (logical_identifier) for test cases (test .xml labels). The current examples in the ldd-template repo use urn:nasa:pds:bundle:collection:product, but it seems useful to encourage stewards to use a different LIDVID for each test case, ideally something meaningful. One idea would be

urn:nasa:pds:namespace_<namespace>:test:test_invalid_value_FAIL

for a test label named test_invalid_valid_FAIL.xml. Other thoughts/suggestions?

2. I recommend including other instructions such as:

• Make test .xml filenames as specific as possible
• Include a comment at the top of the .xml file indicating the purpose of the test
• Include a comment near the line(s) inside the file that cause failure (for tests expected to fail)
• Start with a minimal working label and only introduce content needed to exercise the desired test, to make the test cases easier to read

3. If this discussion results in changes to the guidelines for test cases, we'll want to make sure that the test cases in the ldd-template repo follow the guidelines.
   Relatedly, one of the example test cases provided in the ldd-template repo does not follow the current guidelines (to specify [IM_VERSION] instead of a specific information model version):
   https://github.com/pds-data-dictionaries/ldd-template/blob/main/test/test1_FAIL.xml
   (unless this was deliberate in order to make it fail for other IM versions?). I can't seem to create an issue for the ldd-template repository, so I'm noting it here for @jordanpadams :) .

Have you identified other good practices for test cases that we can/should share and adopt? Please add your comments here.

[jstone-psi]

I've incorporated some of your suggestions into the survey dictionary at (https://github.com/pds-data-dictionaries/ldd-survey/tree/tests/standardize-test-names). I think that you can do a little bit more with the collection names, as well. In this branch, I'm trying data_test_<test_suite> for the collection id.

Additionally, I've added a README.md file to the test directory that specifies what each test does, so you can learn about all of the tests in a single centralized location. Putting a comment near the expected failure would still be a good idea. And so would putting a comment at the top of the file.

[wkiri]

Excellent idea to add a test suite README.md, @jstone-psi !

I also like your data_test_<test_suite> convention (or just test_<test_suite>?).

Btw, it looks like this test is missing an underscore in the collection name, datatest_rollover_fwhm_units:

https://github.com/pds-data-dictionaries/ldd-survey/blob/1d4ae5c2c578a32ddf40a3feff574bc8fb8f336b/test/test_rollover_fwhm_units/test_rollover_fwhm_units_FAIL.xml#L16

[jstone-psi]

Wow, I was sure that the collection type had to be in the collection_id, but I can't find anything that supports that. In that case, test_<test_suite> should be fine.