refactor: add entry-point orchestration and StepRunner - #284
Conversation
079a92a to
668f8ca
Compare
8e95221 to
ca27170
Compare
78273f6 to
857d83c
Compare
857d83c to
ce8abd1
Compare
35d3130 to
7e9e426
Compare
Rewrite entrypoint.py from a stub to a full bootstrap that initialises settings, logging and UI detection, then orchestrates seven action phases (update, migrations, app install, prefix setup, data sync, monitor launch, troubleshooter) via a dependency-gated _run_flow(). Add core/runner.py with the StepRunner class providing guarded step execution (try_step / try_step_code) and higher-level phase wrappers (try_phase / try_phase_code) with automatic result tracking. Suppress ty unresolved-import warnings for deferred module imports that will be implemented in follow-up commits.
7e9e426 to
4766365
Compare
shymega
left a comment
There was a problem hiding this comment.
I'm not comfortable with how much is being done in one MR.
We've also had feedback about our usage of AI, and I think we should focus on a policy to try and address this before adding more LLM-generated code.
I'm not against AI, but I fear that that this commit has had little human input, and AI should assist us, not do it all for us.
At what point do you feel that way? |
|
Basically, the main thing I want to see done differently is a laid-out plan, agreed between both of us, before implementation. That way we're working together, and have a solid agreement on how we want to do this. I'm not necessarily opposed to this PR in fundamentals, but it changes a lot, and should be done in stages. We could have one PR implementing the StepRunner class, then we could agree on the steps we need. I feel a lot of it needs discussing. Sorry for the delay on getting back to you. I've been trying to work this out in my head. |
Shure.
I think I have an idea of how you wana do this. |
Revamp entry_point.py with dependency-gated phase orchestration and extract StepRunner into core/runner.py.
Depends on: #283