Python PostgreSQL modules. Which is best?

Question

I've seen a number of postgresql modules for python like pygresql, pypgsql, psyco. Most of them are Python DB API 2.0 compliant, some are not being actively developed anymore. Which module do you recommend? Why?

Answer 1

psycopg2 seems to be the most popular. I've never had any trouble with it. There's actually a pure Python interface for PostgreSQL too, called bpgsql. I wouldn't recommend it over psycopg2, but it's recently become capable enough to support Django and is useful if you can't compile C modules.

Answer 2

I suggest Psycopg over Psycopg2 since the first one seems a bit more sable. At least in my experience. I have an application running 24/7 and sometimes I was getting random memory crashes (double free or corruption errors) from Psycopg2. Nothing I could have debug fast or easily since it's not a Python error but a C error. I just switched over to Pyscopg and I did not get any crashes after that.

Also, as said in an other post, bpgsql seems a very good alternative. It's stable and easy to use since you don't need to compile it. The only bad side is that library is not threadsafe.

Pygresql seems nice, there's a more direct way to query the database with this library. But I don't know how stable this one is though.

Answer 3

Psycopg1 is known for better performance in heavilyy threaded environments (like web applications) than Psycopg2, although not maintained. Both are well written and rock solid, I'd choose one of these two depending on use case.

Answer 4

In my experience, psycopg2 is the most used library for this. Like you said, it's DB API 2.0 compliant, wich gives a solid interface to work with.

For those who find the standard API to be a little too verbose and hard to work with, I made a small library that might help:

https://github.com/hugollm/rebel

Answer 5

I uses only psycopg2 and had no problems with that.

Answer 6

I've used pg8000 without any problems in the past 3 years. It's uptodate and available on pypi and works on both python2 and python3. You can use "pip install pg8000" to quickly get it (don't forget to use --proxy=yourproxy:yourport if you're behind a firewall).

If you're worried about thread safety, it also provides a score for thread safety (see: http://pybrary.net/pg8000/dbapi.html and https://www.python.org/dev/peps/pep-0249/ for definitions of the different levels of thread safety) (although I have no yet used threads with psql).