Names of tests in modules. <Currently>, we have th...
# team-maintainers
s
Names of tests in modules. Currently, we have this description:
Test names SHOULD describe the test dataset and configuration used. some examples below:
```test("homo_sapiens - [fastq1, fastq2] - bam")
test("sarscov2 - [ cram, crai ] - fasta - fai")
test("Should search for zipped protein hits against a DIAMOND db and return a tab separated output file of hits")```
This is being taken to mean that we should always describe the species, and other generally not very relevant detail. But in very few modules do we actually test multiple species, or filetypes, and so mostly this just because something for reviewers/authors to complain about. In fact I'd say that the majority of modules just have two tests, one with and one without a stub. I'd therefore say that we don't actually care about the names of the tests at all, unless there are more than these two, and then we only really want to capture the important detail.
👍 1
f
I think thats a valid point but I would like to add that sometimes its nice to directly see from the test name what it includes (but I also ported a lot of modules with more than just two tests). Especially if its a tool that is usually applied to human data and then the test is on sarscov2. But I would be ok to lower this specification.
👍 1
s
Almost all the pytest conversions from the last few months have been for modules which either import another module or have more than one test inside them, because Usman's conversion did almost all the simple conversions last year 😉
👍 1
j
I think I would be fine to say 'test names should be sufficiently descriptive'
bu t keep as examples
As I agree with Famke, the test names CAN be very useful when you've got loads - so it does matter
m
I vote for test names should be sufficiently descriptive. Tests name are super useful, so they have to be unique, and self explanatory, so we what is being tested and where
👍 2
s
what is "sufficiently descriptive"?
m
What is life?