Somebody previously suggested this (@MonsieurNicolas maybe?), although I don't remember why we didn't do it then.
I'd like to do it now so it is clearer what is failing when we're updating SDKs, or accepting contributions.
Now that we have each example isolated as its own workspace, running make test fails early on the first example that hits an error. It'd be helpful when looking at PR CI if the failures were across the entire set of examples so we understood how far away a PR is from being ready.
Running in isolation may not be faster as a single run uses a lot of instances for a short period of time. It shouldn't be prohibitively long though.
What
Run each example in isolation in CI
Why
Somebody previously suggested this (@MonsieurNicolas maybe?), although I don't remember why we didn't do it then.
I'd like to do it now so it is clearer what is failing when we're updating SDKs, or accepting contributions.
Now that we have each example isolated as its own workspace, running make test fails early on the first example that hits an error. It'd be helpful when looking at PR CI if the failures were across the entire set of examples so we understood how far away a PR is from being ready.
Running in isolation may not be faster as a single run uses a lot of instances for a short period of time. It shouldn't be prohibitively long though.