Feature request - Allow to use a given instance of EventLoopGroup instead of creating a new one in each thread about pushy HOT 8 CLOSED

Agreed. This is on the to-do list; it came up in #9 and #10, too.

from pushy.

Turns out that doing this will make fixing a few other bugs substantially easier. Moved this up to the v0.2 milestone and opened a pull request (see #18).

from pushy.

@flozano Somebody at @relayrides has pointed out that I misunderstood what you're asking for here. I think #18 is a step in the right direction, but doesn't allow for sharing between push managers (though it does now share between connections within a single push manager). I think I'm going to (re-)defer the shared-among-push-manager aspect of things to v0.3, but will still get #18 in for v0.2. Pardon the confusion!

from pushy.

Actually, just sharing a group between PushManagers isn't so bad. That's done in cd78b1c, so #18 should, indeed, fix this issue.

from pushy.

ah cool :)

I guess loop group lifecycle is handled differently if I give it to the PushManager instance as opposed to an internally managed one... I assume if I give mine to PushManager, PushManager won't kill it when I close it, right?

from pushy.

Nevermind, just saw the commit and the field

• private final boolean shouldShutDownWorkerGroup;

indicates you were already thinking on this 👍

from pushy.

Yeah -- tried to make it clear in the Javadoc, too. If there's anywhere else you think I should advertise it, I'm all ears.

from pushy.

I think it´s pretty natural behavior (if I provide it, I handle - if you have your own, you handle), so I think I would be OK with just mentioning in javadoc - but, of course, I asked for this feature so it´s natural to me.

Thanks for hearing my suggestion btw!

from pushy.