-
Notifications
You must be signed in to change notification settings - Fork 1.3k
SWR cache persistent across tests even with cache clearing and deduping interval at 0 #781
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Comments
Ran into this one again, can confirm that workaround worked for me on a single fetch test which at least is some progress, still I think this needs to have a more substantial fix as it's causing quite a bit of pain. |
Another workaround that I've found: afterEach(async () => {
await waitFor(() => cache.clear())
}) |
Do any planning / instruction for fix it ? i have a and seem
|
Anything new regarding this issue? Had the same problem in my last tests and it's quite difficult a.) to debug and b.) find a solution. At least there should be a hint in the packages documentation. |
We are addressing these issues (better testing support) in #1017, it’s still WIP but any feedback is welcome! |
At the end what is the solution? Implement our own cache and do not cache values when in test environment? |
You can reset the cache per test with let provider
describe('...', async () => {
beforeEach(() => {
provider = new Map()
})
it('....', async () => {
const { cache } = createCache(provider)
useSWR(key, fetcher, { cache })
})
}) In this case, each unit test has its own cache, which won't be effected among all test cases. All the states will be written to the corresponding cache. See related docs: https://swr.vercel.app/docs/advanced/cache. |
For those having issues with RTL and SWR, swr cache.clear() after every test handles a global cache and I think this paired with RTL isn't working always as expected which leads to cached responses to be used across tests. However, the new version of SWR handles cache on tests differently and you can now have a cache for each rendered component separately. So this has fixed for me: previous swr solution (does not clean cache correctly between rtl tests)
new swr 1.0 solution:
This fixed for me as each component has its own cache so there is no more caching issues between tests. |
Uh oh!
There was an error while loading. Please reload this page.
Bug report
Description / Observed Behavior
I have found that some sort of SWR cache(?) persists between tests regardless of clearing the cache and setting the deduping interval to 0. The results are unpredictable and vary each time I run the tests.
Expected Behavior
With cache clearing and deduping interval set to 0, I expected the components using SWR to return the correct data from the API for that test case, not from the previous test case.
Repro Steps / Code Example
Code example here: https://github.com/frdwhite24/swr-testing
Example test script is here: https://github.com/frdwhite24/swr-testing/blob/master/src/component/FetchData.test.js
git clone https://github.com/frdwhite24/swr-testing.git
cd swr-testing
npm install
npm test
(try running this 10 times and watch the different results)Additional Context
SWR v0.3.9
Jest as the testing lib
I have implemented mocking of my API using MSW (mock service worker) and am replacing an existing query handler with a runtime request handler. Starting out with a minimal code base, I tested out the MSW runtime request handler resetting with a vanilla fetch command. After this worked 100% of the time I introduced a component using SWR to get the data, and tests that run before, with and after I implement the runtime req handler. I found that one of three scenarios occur:
I have tried putting a vanilla fetch between the SWR tests and this seems to help it return the correct data.
I have also put two SWR tests after each other which are identical, and the first one will fail but the second one will pass.
I have logged the SWR cache before and after each test and can see that it is being emptied.
I have logged the number of request handlers being used by MSW and have confirmed that they are as expected.
I have also forced MSW to reset the handlers between tests using the specific list of handlers used at the initial state.
I have tried running each test in isolation using
test.only
and each one passes 100% of the time.I've been fighting this one for many days now and appreciate the time you spend looking into it!
The text was updated successfully, but these errors were encountered: