Thanks for fixing the existing tests. Two things:

1. I presume this is a bug on 3.3.x as well as the unreleased 3.4.0. If so, 
could you please re-target this PR at the `tp33` branch
2. You provided a small python script to reproduce that problem. Is there any 
reason you could not convert that into a test as part of this PR so that we 
have some insurance against regression?

> ideally the client (who handle the connection pool) should allocate and free 
> the connections and not the connection itself

I think your approach is fine, but I suppose that the `Client` which borrows 
the `Connection` should be the thing to return it to the pool. If you felt like 
refactoring that at some point I think that approach would be ok as well. 
Thanks.

[ Full content available at: https://github.com/apache/tinkerpop/pull/1013 ]
This message was relayed via gitbox.apache.org for [email protected]

Reply via email to