Async branch of this library?

[smlbiobot]

Hi, great work on this library. I have been using it with lots of success. I am working on an asynchronous project right now, however, and I am wondering if there is an async equivalent of this library.

I am looking into http://aioes.readthedocs.io/ which looks to be an async version of elasticsearch-py, but after using your DSL library and written all my models, I didn’t quite want to go back to the barebones way.

I haven’t read your source, but based on the blocking code:

from elasticsearch import Elasticsearch
from elasticsearch_dsl import Search

client = Elasticsearch()

s = Search(using=client, index="my-index") \
    .filter("term", category="search") \

response = s.execute()

Would it work if I just just plug aioes’ Elasticsearch into it like this:

from aioes import Elasticsearch
from elasticsearch_dsl import Search

client = Elasticsearch()

s = Search(using=client, index="my-index") \
    .filter("term", category="search") \

response = await s.execute()

Alternatively, can you suggest how I might go about this OR if you have plans to write an async version of the same library? Thanks!

[honzakral]

It would not work as a drop in replacement, neither aioes nor the official https://github.com/elastic/elasticsearch-py-async would work. You'd have to create a custom subclass of Search that makes the execute method a coroutine.

[smlbiobot]

ah ok — I didn’t even know about an official async library so I will definitely look into it. Would you consider writing one though?

[digitaldavenyc]

@honzakral I think quite a bit more would be required, simply making Search async wouldn't be enough, there would need to be a full-on Async option.

Does it make sense to build that into the existing library or does it not belong here and another repo should be created to maintain an async version of elasticsearch-dsl?

[smlbiobot]

It would be incredible if you would consider this. Right now I have no choice but to use the aioes library and lost all my Python-ness within the process and had to deal with dictionaries again…

[honzakral]

Closing this as a duplicate of #556