This message was deleted.

We have test suites running in CI that have to pass in order to automate more downstream pipelines. So devs need to keep it green to have the pipeline flow. We need to make better use of codecov to enforce better rules about tests before letting a PR get merged. Also someone with elevated privileges can still force the pipeline forward even if CI is failing tests

Platform Team can have a dedicated service to run automated performance tests or security tests (which are not always closely-related to business logic). It can be either run by the Platform Team on demand of Stream-aligned Teams - or a self-service option

I was thinking about something similar to @Teddy but in relation to PACT testing. Eg failing pipelines if contracts are no longer being honored (as defined by the consumer).

Disclaimer: I’m founder of Qovery @vidhya I think you might like my webinar (a few weeks ago) on this topic. The record is available here (written guide here). I covered how E2E testing can be achieved with Ephemeral Environments with an Internal Developer Platform (Qovery in that case). Still, even if you abstract away Qovery, you might find it quite interesting what’s possible to improve quality products. It’s a use case that dozens of mid-size companies and enterprise platform engineering teams have integrated into their quality process.

The research still says the most effective tests are written by the developers, so the writing of tests should remain with devs. However, platforms can give them better ways to run the tests, as Romaric mentions above. It also may give them ways to incorporate other kinds of testing around security, supply chain, or other concerns that aren't part of the developer tests right now. The toughest problem will be convincing folks to write those tests, though. There is still so many bad takes on how these tests are a waste (and some of the methods picked up from books kinda make it true by encouraging a design that makes tests brittle).