differentiate between infra errors and actual errors

[loosebazooka]

tests in test_expiration require the existence of "faketime" on the system. Perhaps failures on missing infrastructure pieces can be surfaced differently?

[jku]

this case is handled in the Makefile I believe but I suppose we could handle this in pytest as well: just refuse to run without faketime being available

[loosebazooka]

I think my "make dev" succeeded without faketime. Happy if it fails there

[jku]

oh right, it probably only fail when running the included clients (make test-go-tuf) : this is not helpful in your case.

So options would be

• make dev should fail without faketime
• pytest refuses to run if faketime is missing
• individual tests refuse to run if faketime is missing

[loosebazooka]

This might matter a lot less when we're conformant. But getting started is confusing if it errors like this. Local dev feel like critical path for getting started