You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Currently there are two flavours of "runtime" (or "runner" - the actual test runner part) in cucumber-js: serial (the default) and parallel. This has always been one of the harder parts of the codebase to work with, but it's coming to a head now for a couple of reasons:
The parallel implementation has some logic around work assignment that's very challenging to test because it's all coupled to the implementation of the IPC between coordinator and worker
We want to start supporting plugins that relate to things the runner does (like retries), but the boundaries between the coordinator and worker worlds are unclear, and there is quite a lot of duplication between the serial and parallel implementations such that there's no good place to start this effort without getting bogged down
From the discussion in #2357, the consensus is that we'll build a generic runner architecture with the coordinator and worker concepts at the core, but with different adapters for serial (single in-process worker) and parallel (multiple workers with IPC).
This should have no effect on user-observable behaviour, but should greatly accelerate our efforts around plugins, and make the parallel functionality better tested and less risky in general.
The text was updated successfully, but these errors were encountered:
Currently there are two flavours of "runtime" (or "runner" - the actual test runner part) in cucumber-js: serial (the default) and parallel. This has always been one of the harder parts of the codebase to work with, but it's coming to a head now for a couple of reasons:
From the discussion in #2357, the consensus is that we'll build a generic runner architecture with the coordinator and worker concepts at the core, but with different adapters for serial (single in-process worker) and parallel (multiple workers with IPC).
This should have no effect on user-observable behaviour, but should greatly accelerate our efforts around plugins, and make the parallel functionality better tested and less risky in general.
The text was updated successfully, but these errors were encountered: