Ideally: Monolithic, 1 Language, No need to write your own APIs. Just direct access from the View.
For quick and dirty projects I’ve always wondered what a Python version would look like.
Maybe Jinja2 pages, a place to define extra code to be run or a custom ninja tag for longer sections of Python, and a binary to launch the whole thing as a server?
If you e.g. have a use case where your dependency-situation is one where you won't benefit from bottle being single-file (because you e.g. have a ton of other non-single file dependecies), then choosing it over the widely used Flask might not be show you any benefits at all.
For the record, I've used both, and Bottle is much faster to get started with. The docs are also much leaner.
This tutorial is a pretty in-depth guide that incrementally adds more features to a flask app: https://blog.miguelgrinberg.com/post/the-flask-mega-tutorial...
By no means do you need to use everything in here, but I reference it from time to time. This is plenty:
from flask import Flask, render_template
app = Flask(__name__)
@app.route("/")
def index():
return "<h1>hello world<h1>"
@app.route("/foo")
def foo():
return render_template("foo.html", vars={"foo": "bar"})
if __name__ == "__main__":
app.run()You mentioned "weekend projects", which indicates that you value development productivity. You want to be able to do more with your allotted time. The way we tend to envision a "simple stack" is to mean a combination of lighter and possibly specialized tools. But does that necessarily translate into quick development? Your choice of tools still relies on your experience with them individually.
I often see people advocating for microframeworks when this question comes up. As a long time user, I would say sure, but with the caveat that they come with very few batteries, and sometimes fewer opinions. To me this is also the promise of many future decisions to be made, by you. How much time does a single decision typically cost you?
Also, the size of your project's audience, doesn't imply that the app needs considerably less dependencies, than something more ambitious popularity wise. The difference usually lies in the infrastructure that will eventually specifically address scalability.
I think a good simple stack for week-end projects should aim to get you from install to development fairly quickly. Don't waste time making too many technical decisions beyond the choice of tools. Spend half an hour setting up a container, design your database, and get to work. If some other people have already spent time working out what's best in 90% of cases, use that experience first. Find out later that your case falls within the 10% that doesn't apply.
---
Now, for some more practical recommendations. I think Django's opinionated batteries make it a more compelling solution for "week-end" type projects. Does it matter that it installs megabytes on disk, of which you end-up barely using a few KBytes? For most cases I don't think it does.
My stack suggestions for week-end type projects on the back-end will be heavily Python-centric (except for Ruby-On-Rails):
- Django|RoR, Postgres: Opinionated batteries. Just hop on the train and simply let it drive you.
- Flask, Psycopg2, Postgres: when you're already comfortable replacing most things Django offers for free, with your own assortment of best-of-breed tools and practices (testing, templating, etc). Working with Psycopg2 implies that you're reasonably comfortable working directly with SQL, and willing to spend some additional time working out the various kinks that inevitably need addressing, when managing your own connections to the database.
- Flask|Django, Postgres, SQLAlchemy: Only if you already know SQLAlchemy. If you don't, just trust that you won't learn it in a week-end.