Rodolfo Carvalho a22c6bfc8f Fix typo (#6704) 4 anos atrás
..
aiohttp 4cba05ef01 aiohttp improvements (#6663) 4 anos atrás
api_hour 4c7d9415a4 Bump pyyaml from 3.11 to 5.4 in /frameworks/Python/api_hour (#6490) 4 anos atrás
apidaora 0096f84ee9 Config for the new toolset (#6259) 4 anos atrás
async-worker 180c2e1fe3 Adding Python / async-worker framework (#6602) 4 anos atrás
blacksheep 9d18b6e21b make sure all have versus key (#6265) 4 anos atrás
bottle fb8f0c603e Bump bottle from 0.12.13 to 0.12.19 in /frameworks/Python/bottle (#6522) 4 anos atrás
cherrypy 9d18b6e21b make sure all have versus key (#6265) 4 anos atrás
crax 9d18b6e21b make sure all have versus key (#6265) 4 anos atrás
django e33ae563de Bump django from 3.1.5 to 3.1.12 in /frameworks/Python/django (#6629) 4 anos atrás
emmett 28efeb81e2 Update Emmett to v2.2.0 (#6429) 4 anos atrás
eve 0096f84ee9 Config for the new toolset (#6259) 4 anos atrás
falcon 0096f84ee9 Config for the new toolset (#6259) 4 anos atrás
fastapi 0096f84ee9 Config for the new toolset (#6259) 4 anos atrás
flask a22c6bfc8f Fix typo (#6704) 4 anos atrás
hug 0096f84ee9 Config for the new toolset (#6259) 4 anos atrás
japronto 9d18b6e21b make sure all have versus key (#6265) 4 anos atrás
klein 97dfc71da4 Bump jinja2 from 2.10.1 to 2.11.3 in /frameworks/Python/klein (#6470) 4 anos atrás
morepath bb53a07deb Bump jinja2 from 2.10.1 to 2.11.3 in /frameworks/Python/morepath (#6473) 4 anos atrás
pyramid 0096f84ee9 Config for the new toolset (#6259) 4 anos atrás
quart 0096f84ee9 Config for the new toolset (#6259) 4 anos atrás
responder 2fab796ff9 Bump asyncpg from 0.20.1 to 0.21.0 in /frameworks/Python/responder (#6543) 4 anos atrás
sanic 0096f84ee9 Config for the new toolset (#6259) 4 anos atrás
spyne 0096f84ee9 Config for the new toolset (#6259) 4 anos atrás
starlette 7b928fe504 Bump asyncpg from 0.20.1 to 0.21.0 in /frameworks/Python/starlette (#6544) 4 anos atrás
tornado 0096f84ee9 Config for the new toolset (#6259) 4 anos atrás
turbogears 84db92b766 Bump jinja2 from 2.10.1 to 2.11.3 in /frameworks/Python/turbogears (#6469) 4 anos atrás
uvicorn 67c7a0df02 Bump asyncpg from 0.20.1 to 0.21.0 in /frameworks/Python/uvicorn (#6545) 4 anos atrás
uwsgi 0096f84ee9 Config for the new toolset (#6259) 4 anos atrás
vibora 9d18b6e21b make sure all have versus key (#6265) 4 anos atrás
web2py 0096f84ee9 Config for the new toolset (#6259) 4 anos atrás
webware 9d18b6e21b make sure all have versus key (#6265) 4 anos atrás
weppy 0096f84ee9 Config for the new toolset (#6259) 4 anos atrás
wsgi 0096f84ee9 Config for the new toolset (#6259) 4 anos atrás
README.md 4322cacaef Update documentation links (#4541) 6 anos atrás

README.md

Python frameworks

The information below contains information specific to Python. For further guidance, review the documentation.

Infrastructure Software Versions

Get Help

Python Experts

  • INADA Naoki (@methane) -- CPython core developer and expert of Python's MySQL driver.
  • Ludovic Gasc (@GMLudo) -- AsyncIO and aiohttp user.

Python Community

Python interpreters

(C)Python 3

Newest Python.

It is the most major interpreter for Web Development.

(C)Python 2

Legacy Python. Still used for widely.

PyPy3

PyPy is the fastest Python implementation with JIT.

PyPy2

Legacy PyPy. Still used for widely.

WSGI Servers

Nginx + uWSGI

This is one of fast and realistic way to serve Python web application.

Use unix domain socket between nginx and uWSGI to avoid additional TCP/IP overhead.

Gunicorn + Meinheld

Meinheld is very fast WSGI server.

Since Meinheld is safe against slowloris and support Keep-Alive, you can use it without buffered HTTP reverse proxy (like nginx).

We use meinheld to measure bare (without DB access) performance of framework without overhead of reverse proxying. (plaintext, json test)

Meinheld does not provide worker process management. Gunicorn provide it for Meinheld.

Gunicorn + Tornado

uWSGI + PyPy is difficult to setup. Meinheld doesn't so fast with PyPy because it uses Python/C API heavily. So we use Tornado as HTTP/WSGI server.

It supports keep-alive. So it have nice performance about json or plaintext benchmark.

Before writing new tests.

Don't increase matrix without significant purpose.

We can't maintain matrix of frameworks (5~) * servers (3~) * DBs (3) * Interpreters (3).

If you want to know about performance difference between MySQL and PostgreSQL, there are no need to all frameworks implement both of them.

Additionally, Python is not a good language to measure performance of DBs because web application written in Python is slower than DBs.

If you want to benchmark http server, you should not port all tests or frameworks. Simple json or plaintext test is enough.

If your framework uses SQLAlchemy, Flask may be enough to know performance of SQLAlchemy.

Advice when writing new tests

Interpreter

Consider Python 3 first. -- Python 3 is mainstream for web development.