mirror of
https://github.com/enso-org/enso.git
synced 2024-12-26 12:21:31 +03:00
354ee94a2f
- As asked for by @hubertp who was encountering flaky test failures on CI in the Http_Spec and related ones, I'm adding retry logic to make such cases much less likely. - I've made the test server randomly fail 50% of tests and with the retry logic the tests are still passing, so I think that should be much more robust, in practice the failure rate is much much less (I imagine <1% as most of the time these tests were working and we do a ton of requests in a single CI run). - I move the `with_retries` method to now be `Test.with_retries` which can be used anywhere in our tests for the retry logic. - It sleeps for 0.1s between retries. Not all kinds of tests need it, this was mostly for propagation delays in the Cloud in our tests. I was thinking if the delay should be configurable, but I think the 0.1s delay is not problematic and if our tests are sometimes failing due to high machine load, the delay could also help. - This _does not_ add retry logic to raw HTTP operations or `Data.fetch`. We may add that later, but that needs some further design. In such case we may remove some retries from tests if they become unnecessary. |
||
|---|---|---|
| .. | ||
| bin | ||
| engine/THIRD-PARTY | ||
| launcher | ||
| lib/Standard | ||
| project-manager/THIRD-PARTY | ||
| enso.bundle.template | ||
| launcher-manifest.yaml | ||
| manifest.template.yaml | ||