[Python Dependency Pitfalls] Artisanal "from-scratch" development

Hey there,

The other day I got this question from Newsletter member Newvick in my email inbox:

~~~

I'm trying to get past the beginner's stage in Python and one problem I have is:

When do you use libraries/frameworks and when do you make things from scratch? (whether for learning or developing something for a client/work)

~~~

When to code functionality from scratch and when to use libraries, that's an important question.

There's a fine balance you need to strike here:

When I learned how to program I was "reinventing wheels" left and right—and it pushed my abilities forward quickly.

I was constantly writing all kinds of existing functionality from scratch and it was the right thing to do because it helped me grow my skills.

It was a great learning exercise.

But there comes a time in a developer's life when that strategy doesn't work so well any more:

Eventually I found that I wasn't learning as quickly anymore trying to build everything from scratch.

Plus my development pace was dog-slow.

I was spinning my wheels—and my productivity began to plateau.

Another downside is this:

It's difficult to work on a team when someone always wants to approach every single problem from first principles. It can drive coworkers and managers NUTS.

These days I try to use (good quality + well-documented!) third-party libraries where possible.

But again, you'll want to find a balance there:

For example, any professional chef worth their salt needs to learn how to make mayonnaise from scratch.

But eventually they'll know how to prepare it—and what great quality looks like.

From this point on buying pre-made mayo off the shelf can be a huge time saver:

In some dishes it simply WON'T make a difference on the end result.

It just *takes longer* to make everything from scratch…

Bringing it back to Python dependency management, here's my general rule of thumb:

"Only consider 'write from scratch' if time is NOT an issue. Otherwise use 3rd-party libraries and frameworks—but don't overdo it and only go for the trustworthy ones."

(Remember, you'll want to avoid another "left-pad" incident.)

Sometimes it's just tough as nails to strike the right balance here…and there's so much more to talk about on this topic.

If you want to dig deep and see how to make the leap from "amateur cook" to "professional chef" as a Pythonista, my Managing Python Dependencies course shows you how to get there.

It includes a complete step-by-step workflow for researching packages and making Python dependency decisions (and explaining them to your team, manager, or client).

Click the link below to learn more:

realpython.com/products/managing-python-dependencies/

— Dan Bader

P.S. "Re-inventing the wheel" disease was really something I struggled with as a fledgling dev. I'll show you how I got over it tomorrow.