build VS psycopg2

If it's pure Python, the only packaging file you need is `pyproject.toml`. You can fill that file with packaging metadata per PEP 518 and PEP 621, including using modern build tooling like flit[1] for the build backend and build[2] for the frontend.

With that, you entire package build (for all distribution types) should be reducible to `python -m build`. Here's an example of a full project doing everything with just `pyproject.toml`[3] (FD: my project).

[1]: https://github.com/pypa/flit

[2]: https://github.com/pypa/build

[3]: https://github.com/pypa/pip-audit

RE the dislike of a "third-party tool", what do you mean by this? All the major tools for packaging in Python are under the PyPA, e.g. - twine - build - hatch

building wheels/sdists to upload or install: build

python3 setup.py sdist bdist_wheel - no need for an sdist when building wheels, also I'd recommend using pypa/build instead.

The first step is to turn the internal package into a wheel (a *.whl file). We can use the build tool for this purpose. After installing build with pip we can run it as follows:

Shouldn't https://github.com/pypa/build help? I don't think it has enough features yet but I was under the impression that was the distro solution.

You can practice both by using https://www.psycopg.org from your Python code to communicate with your database. When I wanted to practice some SQL, that's what I did (we use psycopg at work, so that's what I practiced with, making a dream journal thingy for myself that was better than just noting stuff in a notepad because I could then look up e.g. what other stuff was correlated with Y, how many times I dreamed of X, etc. etc.)

But seems like it should work with Django 3, which you have specified https://github.com/psycopg/psycopg2/issues/1293

If you're using Postgres for your SQL, look at the "copy' method of the psycopg module (see https://www.psycopg.org/articles/2020/11/15/psycopg3-copy/) . It's much faster than INSERTs in my experience (YMMV).

Back when I used Psycopg2, there was no -binary package, so you'd get libpq set up similarly to pg-native. Docs say:

> The binary package is a practical choice for development and testing but in production it is advised to use the package built from sources.

Relevant GitHub discussion: https://github.com/psycopg/psycopg2/issues/674

I dunno, this seems worse to me.

Just go to the more easily readable docs here. I’m sorry, but the linked article is terrible.

psycopg2-binary. See https://github.com/psycopg/psycopg2/issues/1286.

psycopg2-binary, PostgreSQL Database adapter for python.

... # psycopg2==2.9.3 # https://github.com/psycopg/psycopg2 ...

I sometimes do this in testing. For example, consider the library used to communicate with a Postgres database, psycopg.