Open limonte opened 2 years ago
This is a mock. It's really weird for it to take that long.
This is a mock. It's really weird for it to take that long.
In this case, it might be something else than mock. By "locally" I meant the actual extension, not the testing environment. So, the actual https://www.googleapis.com/gmail/v1/users/me/settings/sendAs
took 12 seconds once and I'm pretty sure the Internet connection was stable.
This issue might be tricky to catch because it's flaky and happens quite rarely.
For the CI, are you referring to this?
✔ setup - import key - fix uids 00:56
ℹ (attempt 1 of 1) Failed: TimeoutError: waiting for selector `[data-test="input-compatibility-fix-expire-years"]` failed: timeout 20000ms exceeded 00:56
at new WaitTask (/home/semaphore/git/flowcrypt-browser/node_modules/puppeteer/src/common/DOMWorld.ts:798:28) 00:56
at DOMWorld.waitForSelectorInPage (/home/semaphore/git/flowcrypt-browser/node_modules/puppeteer/src/common/DOMWorld.ts:652:22) 00:56
at Object.internalHandler.waitFor (/home/semaphore/git/flowcrypt-browser/node_modules/puppeteer/src/common/QueryHandler.ts:78:19) 00:56
at DOMWorld.waitForSelector (/home/semaphore/git/flowcrypt-browser/node_modules/puppeteer/src/common/DOMWorld.ts:510:25) 00:56
at Frame.waitForSelector (/home/semaphore/git/flowcrypt-browser/node_modules/puppeteer/src/common/FrameManager.ts:1274:47) 00:56
at Page.waitForSelector (/home/semaphore/git/flowcrypt-browser/node_modules/puppeteer/src/common/Page.ts:3210:29) 00:56
at ControllablePage.ControllableBase.waitAll (/home/semaphore/git/flowcrypt-browser/test/source/browser/controllable.ts:57:27) 00:56
at Function.manualEnter (/home/semaphore/git/flowcrypt-browser/test/source/tests/page-recipe/setup-page-recipe.ts:189:28) 00:56
at /home/semaphore/git/flowcrypt-browser/test/source/tests/setup.ts:110:7 00:56
at /home/semaphore/git/flowcrypt-browser/test/source/test.ts:76:7 00:56
at <async> waitAll (/test/source/browser/controllable.ts:46:19) 00:56
at <async> manualEnter (/test/source/tests/page-recipe/setup-page-recipe.ts:87:4) 00:56
at <async> <anonymous> (/test/source/tests/setup.ts:108:76) 00:56
at <async> <anonymous> (../test/source/test.ts:71:53) 00:56
at <async> <anonymous> (/test/source/browser/browser-pool.ts:143:49)
(if not, please point me to the right direction)
If yes, then these are two different issues between the CI and your local environment.
setup - import key - fix uids
is notoriously slow. I don't know what is it about that particular test, maybe it's the fixing of the openpgp key that takes so long, but really, it shouldn't. I never debugged it.For the CI, are you referring to this?
Yes.
Yup - that's another issue.
Gmail API may be something that we can't do much about. On the other hand, speeding up setup - import key - fix uids
could be done, possibly. But I have a feeling the problem is in the extension implementation, not the test.
There was flakiness in tests which appears to be caused by the actual slowness of the app:
The reason is that the
https://www.googleapis.com/gmail/v1/users/me/settings/sendAs
request sometimes takes quite a white, I personally experienced 12 seconds locally, but in CI it sometimes takes more than 20 seconds, example: https://flowcrypt.semaphoreci.com/jobs/9f63948c-b6c9-4123-8d67-236e1985c5b1Not sure if this is something that could/should be fixed, reporting for the reference mostly.