Doh! Yesterday, I was rather content with the decision to use a different driver from Semiocast[1], but now I guess I actually have to evaluate this new epgsql work to see how it compares :(. (Yeah, I imagine this is a rather awkward kind of response to this kind of work ;P... like, I'm glad to see PostgreSQL on Erlang getting some love, but at least with epgsql in the "semi-dead" state it was easy to not have to evaluate it ;P.)

[1]: https://github.com/semiocast/pgsql

(I have now posted this comparison question to the new mailing list[2]. I must say it is incredibly refreshing to see someone appreciate the importance of mailing lists over the "fork it in public, make some changes, and likely never tell anyone: my work here is done, if they want to merge it I guess that's now their job" style of interaction that GitHub seems to be training the new generation of open source programmers :D.)

[2]: https://groups.google.com/forum/#!topic/epgsql/gd5-NWib_iY

Yeah, I'm familiar with that other driver, and it looks like good work, so you should be ok with it! We were already using the 'mabrek' fork of epgsql in Chicago Boss, which is why it was worthwhile to invest some time in the project. I'm not honestly sure how the two compare, but I know we're not the only ones invested in the epgsql codebase.

I prefer the semiocast driver myself and use it in production at Heroku with no issues. And luckily they have become responsive to issues and PRs. They initially released it when I attempted to do the same thing OP did a year ago. But I instead switched to their client when they released it and I haven't looked back since.

Yes, coordination is important, too. Well done.