Hate to be w downer but I’d never consider this for use.
For trivial applications maybe it’d work, but with more complexity like anything that has risk of deadlocking or depends on the database shape and such solution subtracts from value as even small shift in behavior can snowball into critical problems.
Today I lean towards resource constrained E2E environment so that local test runners have opportunity to break if someone write anything grossly underperforming.
Not to mention that snapshotting DB after second and distributing this snapshot to test partitions is super fast and many times shaved multiple minutes from test suites.
It’s an interesting idea and definitely great learning experience but I think that target audience is limited.
For trivial applications maybe it’d work, but with more complexity like anything that has risk of deadlocking or depends on the database shape and such solution subtracts from value as even small shift in behavior can snowball into critical problems.
Today I lean towards resource constrained E2E environment so that local test runners have opportunity to break if someone write anything grossly underperforming.
Not to mention that snapshotting DB after second and distributing this snapshot to test partitions is super fast and many times shaved multiple minutes from test suites.
It’s an interesting idea and definitely great learning experience but I think that target audience is limited.