Using async-http as a transport

[ioquatix]

Async::HTTP is an asynchronous persistent and scalable HTTP client which works very well within the Falcon server.

I have been reading https://github.com/elastic/elasticsearch-ruby/blob/master/elasticsearch-transport/README.md#transport-implementations but it's not clear how you achieve thread safety.

Do you use one faraday connection instance per client?

Are you aware of the new hooks to manage life-cycle of persistent connections? e.g. Faraday::Connection#close?

Do you expose those methods from the client?

Just wondering how best to support elasticsearch.

[picandocodigo]

Hi @ioquatix,
I'm the new maintainer for this project and I'm currently working on several things. But I have added this query to my list and I'll get back to you as soon as possible.

Thank you,
Cheers!

[ioquatix]

Ping.

[ajjahn]

Ping.

[ioquatix]

Pong.

[picandocodigo]

Hi @ioquatix, I've transferred this issue to the new home for elastic-transport. You can see how connections are handled in base.rb and the different implementatiosn in http. You can write a new implementation by including Elastic::Transport::Base and implementing the API.

Adding support for Async::HTTP is not on our roadmap at the moment, but we'll gladly review the code if you submit a PR.
Thank you!