Tests should be refactored

[almet]

Some tests would benefit from a refactor. I'm mostly thinking about the logic contained in the _get_or_create private method, which is not directly tested but via its callers.

One solution here would be to create a base class with the behaviours we want to test and then derive multiple classes with the different inputs (bucket, collection and record).

[Sayli-Karnik]

I could work on this..So BucketTest, CollectionTest ,RecordTest must extend a common base class. And are we just dealing with _get_or_create or all methods? @almet @Natim

[sahildua2305]

@Sayli-Karnik Can I take this up, in case you're not?

@almet @Natim Can I have more information about this issue?

[leplatrem]

To be honest I'm not very convinced by the proposition of a «base class», but we can try to see how it goes.

The idea is to go through every buckets/collections/groups/records client tests and see how much code is «repeated» between them. If there is a lot, let's go with the base class approach.

Let us know what you find out ;)

[zeddmaxx]

Hi all
This looks very interesting.
I would like to contribute to this, I hope I can.

[almet]

Well, if you get the thumbs up from @sahildua2305 , I believe you can go ahead and take it!

[sahildua2305]

Sure @zeddmaxx! Go ahead. It's all yours.

[sahildua2305]

@zeddmaxx Any progress on this?

[sahildua2305]

Accounting to no response from you, @zeddmaxx, I'd like to claim it back.

/cc @almet @leplatrem

[sahildua2305]

@glasserc thanks!

Btw, do we want to see how much code is repeated between corresponding tests for buckets/collections/records/groups? Or separately for Bucket tests and then for collection tests and so on?

[glasserc]

I think @almet's original goal was to share code that corresponds to the same test in buckets, collections, records, and groups. But any kind of DRYing up or refactoring could potentially be beneficial.

[sahildua2305]

@glasserc I noticed that there are a lot of tests in Buckets / Collections / Records / Groups which test the same functionality. For example - there are tests to verify that create_xxx and update_xxx issues put http request and that patch issues patch http request.

And that I think is intentional and necessary to make sure we have good test coverage. Making a base class and calling these tests indirectly using that class methods and just passing params necessary for new tests will make tests difficult to understand for new contributors.

What do you think?

[glasserc]

I don't have a strong feeling about it, but I'm not as knowledgeable in this area as @leplatrem and @almet. If we think it's a bad idea, let's close the issue.