Adds a new linter to verify all test names are consistent with one common
convention we already follow, for the most part, which is:
> Test vars (test names) should be suffixed with -test.
There's no strong reason for following this convention, although it's quite
common in Clojure, but in any case, these are the reasons I can think of and
remember:
- Naming consistency. Sometimes tests start with "test-", others end with "test"
and others don't prefix/suffix at all.
- The suffix removes potential conflicts with core Clojure functions.
- The suffix mostly eliminates potential conflicts with other vars in the test
namespace. Example: you can declare a function delete and have a test named
delete-test.
- For someone using Emacs imenu feature, it helps differentiate which vars are
tests and which are just local functions supporting the tests.
Fix all component tests after the latest RN upgrade.
Fixes https://github.com/status-im/status-mobile/issues/18157
Closes https://github.com/status-im/status-mobile/pull/18235
Dependency changes
- Upgraded Jest: from 26.6.3 to latest 29.7.0.
- Upgraded @testing-library/jest-native: from 5.3.0 to latest 5.4.3
- Upgraded @testing-library/react-native: from 11.5.4 to 12.4.2
- Removed explicit dependency on jest-circus, this is now the default test
runner.
- Removed explicit dependency on jest-environment-node. This is handled by the
package manager.
- Added jest-silent-reporter at version 0.5.0.
### Why component tests were failing?
Many tests were failing because we were using RN Testing Library (RNTL) in an
unreliable fashion. With the recent library upgrades, the unreliability was
excerbated. Other times, the tests were incorrectly arranging data.
### with-redefs does not work with async code
Generally speaking, with-redefs should not be used with async code, assume the
worst. The scope of the macro will cease to exist by the time the async code
runs. In many tests we were using with-redefs, then calling render, but for some
components that use use-effect, JS timers, animations, etc it's unreliable and
were the reason for failures.
It's easy to reproduce too:
```clojure
(defn foo []
:foo)
(foo)
;; => :foo
(with-redefs [foo (constantly :bar)]
(foo))
;; => :bar
(js/setTimeout
(fn []
(tap> [:calling-foo (foo)]))
100)
;; Taps [:calling-foo :foo]
;; As you would expect, when running without with-redefs, it prints :foo.
;; So far so good, but whatch what happens with async code:
(with-redefs [foo (constantly :bar)]
(js/setTimeout
(fn []
(tap> [:calling-foo (foo)]))
100))
;; Taps [:calling-foo :foo]
;; ====> PROBLEM: Taps :foo, not :bar as one might expect
```
### Not waiting on wait-for
When test-helpers.component/wait-for is used, subsequent assertions/etc should
be done after the promise returned by wait-for is resolved. But remember to not
perform side-effects inside the wait-for callback (check out the docs
https://callstack.github.io/react-native-testing-library/docs/api#waitfor).
Most, if not all of our usages of wait-for were not waiting.
#### Improvement 1 - Silence Jest on demand
If you need to re-run component tests frequently, you may want to reduce the
output verbosity. By passing JEST_USE_SILENT_REPORTER=true to make
component-test or make component-test-watch you will see a lot less noise and be
able to focus on what really matters to you.
#### Improvement 2 - Selectively focus/disable tests
Because of our need to first compile CLJS to JS before running tests via Jest,
we couldn't easily skip or focus on specific tests. From this commit onwards, we
should never again have to change the list of requires in files core_spec.cljs.
Commenting out required namespaces gives a bad DX because it causes constant
rebasing issues.
#### Improvement 3 - Translations now work as in prod code (but only English)
Translations performed by *-by-translation-text can be done now without any
workaround under the hood. The query functions are now linted just like
i18n/label, which means static translation keywords must be qualified with :t/,
which is good for consistency.
This commit adds a custom linter to verify i18n/label is called with a qualified
keyword, like :t/foo. More sophisticated linters are possible too.
We also set the stage for other developers to consider more lint automation
instead of manually reviewing conventions in PRs.
If you want to understand how to write custom linters, check out
https://github.com/clj-kondo/clj-kondo/blob/master/doc/hooks.md. You can fire
the Clojure JVM REPL in status-mobile and play with the clj-kondo hook too, it
works beautifully.
Why do we care? By making sure all translation keywords are qualified with "t",
it is trivial to grep or replace them because they're unique in the repo, and
can't be confused with other words if you search by ":t/<something>".
Note: It's a best practice to commit clj-kondo configuration from external
libraries in the .clj-kondo directory. The directory .clj-kondo/babashka is
auto-generated, that's why it was added.
Icaro Motta
Dmitri Akatov