pablobm left a comment (openstreetmap/openstreetmap-website#6497)
100% agree with leaving controller/integration tests for low-level testing. My
experience is that these are typically used for testing specific technical
details, variants of "happy paths" reflected in system tests, and malicious
usage of APIs.
My main experience is also with RSpec with the syntax shared by @gravitystorm.
I want to think that with minitest it can be done with something like
`driven_by :rack_test` similar to how there are `driven_by` declarations on
https://github.com/openstreetmap/openstreetmap-website/blob/master/test/system/embed_test.rb,
but I haven't gone that far into testing this idea yet.
Re: Turbo, yes, that's a tradeoff. With rack-test, links and form submissions
will always trigger full-page loads. This creates a difference between what the
tests do and real users do. At this point the question is: how much do we trust
Rails to do what it promises to do?
When using an external tool, we want to think that we don't need to test the
tool itself, but just the things we do with the tool, if that makes sense.
Having said that, things can go wrong for example if something is
misconfigured, etc. Eg: if we don't label a turbo frame correctly by mistake.
Still we can go on a case-by-case basis. For example using Selenium only for
the happy paths, or for specific situations where there's a known risk.
In exchange, we gain a faster execution of the test suite, a reduction of flaky
tests, and more easily debuggable system tests.
--
Reply to this email directly or view it on GitHub:
https://github.com/openstreetmap/openstreetmap-website/pull/6497#issuecomment-3485319404
You are receiving this because you are subscribed to this thread.
Message ID:
<openstreetmap/openstreetmap-website/pull/6497/[email protected]>
_______________________________________________
rails-dev mailing list
[email protected]
https://lists.openstreetmap.org/listinfo/rails-dev
