I'm especially interested in setup, pooling and teardown of a database connection. My prefered example would be postgresql via psycopg2.

It's very important.

For what it is worth: I am using a global variable for my "raw" database connection with hug in

1. https://github.com/Intevation/intelmq-mailgen/blob/master/extras/checkticket-spa/checkticket.py
2. https://github.com/Intevation/intelmq-mailgen/tree/master/extras/contactdb_api

Both uses are still "experimental" and there are a number of drawbacks for using global variables,
one is that I don't know how scaling to different tasks or processes will be handled. My use case for the two examples so far is okay with just a single hug running.

Note that #405 (hug.startup() is not called for cli applications).

A good path forward would be to see how other falcon applications handle such resources that would need to be kept once per serving instance/task or once per process.

Would anyone be interested in a tutorial on how to set up Hug with mongodb, Marshmallow, and pymongo?

@FFX01 more examples are always good I believe, so thanks for offering to write a tutorial. Personally I have already outlined above what interests me most. (How to keep a resource like a psycopg2.pool
object between several threads.)

@bernhardreiter

Unfortunately, I do not have a solution for that use case. I don't know all that much about the psycopg2 internals.