Closed mdivanova closed 4 years ago
How thoroughly have these changes been tested? Ideally they would be (manually) tested by setting up a fresh project, running all make commands, and then running the frontend e2e tests.
I did manually set up a fresh project on cloud shell and run all the make commands. Manual testing of the site looked good.
The frontend e2e tests are flaky though - a different number fail on different runs. The failure errors are not manually reproducible on the site (e.g. Not being able to find a "Create" button on the Items page). I don't think this is related to my refactoring. Are these reliably run by others?
How thoroughly have these changes been tested? Ideally they would be (manually) tested by setting up a fresh project, running all make commands, and then running the frontend e2e tests.
I did manually set up a fresh project on cloud shell and run all the make commands. Manual testing of the site looked good.
The frontend e2e tests are flaky though - a different number fail on different runs. The failure errors are not manually reproducible on the site (e.g. Not being able to find a "Create" button on the Items page). I don't think this is related to my refactoring. Are these reliably run by others?
@zsxking can you comment on e2e test flakiness? Is this expected or should we be concerned that these changes broke something?
How thoroughly have these changes been tested? Ideally they would be (manually) tested by setting up a fresh project, running all make commands, and then running the frontend e2e tests.
I did manually set up a fresh project on cloud shell and run all the make commands. Manual testing of the site looked good. The frontend e2e tests are flaky though - a different number fail on different runs. The failure errors are not manually reproducible on the site (e.g. Not being able to find a "Create" button on the Items page). I don't think this is related to my refactoring. Are these reliably run by others?
@zsxking can you comment on e2e test flakiness? Is this expected or should we be concerned that these changes broke something?
I see the same behavior when running e2e tests on my project from last week's bugash.
The flakiness is expected because of the arbitrary waiting. As long as it's not failing on the same test over and over again, it's probably fine.
It looks like you'll need to change these lines in the PR checks to use the new directory structure