I made a test in `PoolingRestClientTests` that completely recreates client for 
each of 1000 requests, recreating the whole thing seems to add <2ms to requests 
on avg, in my machine. So let's say this would add ~2ms latency to unpause 
workflow (at first run on a paused container). 

I think this is tolerable - WDYT?

I wanted to test with the `pool.shutdown()` approach as well, but after your 
hint above, I'm then stuck at wondering how to deal with the signature on the 
flow either being `Any` or `Http.HostConnectionPool`.  With `Any` on the flow 
sig, the connectionPool ends up as `Any` type, and with 
`Http.HostConnectionPool` on the flow sig, the connectionPool ends up as the 
same type, but I'm not sure how to mock `Http.HostConnectionPool` in the test 
flows (created in test classes). 
Maybe `Any` type is ok here, but will require some casting - is there some 
better way?

[ Full content available at: 
https://github.com/apache/incubator-openwhisk/pull/3976 ]
This message was relayed via gitbox.apache.org for [email protected]

Reply via email to