Thomas B

The current state of #3288 looks good to me (almost) for merge, and the state of the investigation on this issue is such that I'm fine with 2.0 release at...

> Those links are broken for me, or at least they do not jump to any line behavior for me on Firefox and Chromium is that they end up jumping...

It's generally better to not complexify general code with test code so I would tend to not favor this approach if possible. How about an approach where we would create...

> I do not claim that the presented solution is the best way to solve this issue. Everything that follows is just trying to collect some potential problems with other...

> I'm not sure if that solves the problem outlined in #3252, as that specifically would require to leave the connection open even if it's checked in to the pool...

Oh I did not expect there to be so much discussion on that topic already! Funny how we had started the same exact implementation, though I iterated a bit more....

> Also, please squash the commits. We could merge as "Squash and merge". That enables us to keep the history and discussions in the PR, and results in a pretty...

Currently we've been using an `insert_into` `on_conflict` `do_update` to workaround this not being implemented.

Currently we've been using an `insert_into` `on_conflict` `do_update` to workaround this not being implemented.

Also seems to be a duplicate of #1478