SocketIO.run() add additional reloader specific parameters.

[indietyp]

Currently, the SocketIO.run() method allows for some modifications of the reloader, specifically the ability to include extra_files, this pull request adds two new parameters to the SocketIO.run() method to further modify the behavior.

The exclude_patterns and reloader_type parameter, are just redirected to the reloader, but give SocketIO users the ability to further customize how the reloader functions.
This is helpful for projects that have huge e.g. asset folders in their project they would like to exclude, or where a specific reloader just does not work.

[miguelgrinberg]

I think the idea is good, but I'm not sure there is a reasonable way to implement this.

The socketio.run() method supports three different web servers. The solution that you created only works for gevent and eventlet, but is broken for the Flask dev server, which would ignore these arguments. This isn't your fault, the extra_files argument is already suffering from this problem, and I don't feel it is a good idea to make the problem even worse by including more options in this broken solution. Your solution also ignores a few more arguments that you can provide, including reloader_interval which is directly related to the two that you implemented in this PR.

I'm more inclined to say that if you want the full functionality of the reloader, then you should use the Flask dev server, and if you use a different server, then you get a minimal reloader. Supporting the reloader in a gevent/eventlet web server was just a measure that was necessary when the Flask dev server could not support WebSocket connections. This is now possible, so really during development there is no reason to use eventlet/gevent anymore.

[indietyp]

Thank you very much for the feedback!
I have completely overlooked the reloader_interval and other arguments, my bad, sorry - in an ideal implementation all of the arguments would be supported, but I don't think that that is a realistic approach.

Currently all implementations (except the threading based one) construct a run_server() method, that is then called - this could easily be simulated for threading. What would you think about the externalization of the run_server() function creation into a separate method, that way people could still hook into the use_reloader with their own arguments and the current behaviour would not change.

What I propose:
Create a new method (something like _create_run_server()) which returns a callable, run() calls this method to retrieve the run_server() callable and then would call it, depending on the context with a reloader or not. People who would like to further modify the behavior could call _create_run_server() and then create the reloader themselves.

I know that that might not be the best solution, but thought it might the "best of both worlds". What do you think?

Thank you very much for your time and I would be willing to implement such a change - if desired.

[miguelgrinberg]

@indietyp Please have a look at the main branch and let me know if this works for you. I've added a reloader_options that replaces the extra_files argument. This is a dict in which you can add any arguments that are sent to the run_simple function in Werkzeug.

[indietyp]

@miguelgrinberg yes that works for me. Thanks for adding the feature!