How to run FL code after updating Flower source code?

*This question was migrated from Github Discussions.

Original Question:
I want to change the aggregation strategy based on asynchronous updates. For that, I am trying to update the server code. I wanted to know how to test the framework(updated one) while running the client.py & server.py?

Should I build the code using ./dev/build.sh and then install flwr***.whl from dist folder manually? After that I can run client & server code to test the updated framework code. Or do we have any other script that I am missing.

Answer:
Exciting direction!

There’s no need to manually build Flower if the only thing you’re trying to do is to use a custom Server. Both start_server and start_simulation support passing a custom Server implementation:

But - to add to that - if you want to make changes that require a manual build, then yes, you can run ./dev/build.sh and then install the resulting .whl from the dist directory, but you can also just run pip install . in your modified Flower directory.