Software Engineers Hate Code | Prime Reacts

Recorded live on twitch, GET IN

Author: Dan Cowell

MY MAIN YT CHANNEL: Has well edited engineering videos

Discord

[AdroSlice]

Alright boys, hire me. I actually love refactoring code. Because I love pretending like I'm better than the original author.

[danielvaughn]

The difference between writing code and maintaining it, is the difference between raising your hands and keeping them in the air indefinitely.

[TheRustyCrab]

I hate code so much I use chatgippity to generate procedural macros that generate my code for me. Tom is a frickin genius.

[JordanShurmer]

The typography of this article is fantastic. so easy to read.

[richardneumann]

Working as a software dev for almost ten years, I concur, that other peoples' code is the worst. Especially the code of your past self.

[doctorstal]

Actually it's a common trap.
A lot of us had programming as a hobby, while growing up.
And when we take that hobby as a job - it turns out they pay not for writing the code, but for f*ing reading it.
By the time people understand it - it is too late to choose another profession.

[Jmcgee]

9:29 I am always surprised that people can't use the command line effectively, even for basic stuff. Get some basic knowledge of grep, sort, uniq, cut, head, and tail and you'll be able to do a surprising amount of stuff without having to touch more "complicated" tools like awk.

[TCH]

6:49 a competition on who creates the least amount of code for a feature is actually a great idea.

[sciverzero]

Something no one ever teaches about programming... that I think we should establish as an important thing to keep in mind when teaching and when learning...
Coding is easier to _DO_ than to _understand._
It is practically effortless to scream some code into a file, test it and revise until it works...

But its a lot harder to read and understand what existing code does. Everyone approaches a problem differently, and if whoever coded what you have to now read, than what your current state of mind thinks is the best or easiest path to achieving what the code does, you aren't going to understand it without time and effort.

Reading someone else's code, or even your own code from a year ago, is like _learning_ to read, not simply reading.
You may know the symbols and variables and know how to speak them, but you don't know what to expect about how they will be written down, until you completely immerse yourself in how someone has written them down, and eventually you _may_ understand it.

The very next project you look at will require you to relearn everything, because its still not speaking your language, its just using the same words that are in your language.
Reading code is 100% of the time, like knowing Italian, and trying to figure out what something in Spanish says. Sometimes you'll get pieces of it quickly and other times you'll be baffled, and no time will you ever expect to just know what it says or does. You have a basis to understand it, but no reasonable expectation that you will.

There's also no code that works correctly the first time. You code, you test, you fix, and then you have code you can use.
When you're reading code, you don't get to make a small amount of progress, _do a test to confirm_ and then make more. The entire process is just you, acting on faith, that the next thing you read will make some sense based on what you think the last thing you read meant.

[SimGunther]

_Painter hates paint but loves painting_

[AJSquirrel]

The company I work for solved this years ago. No pull requests. No code reviews. Just push straight to production 😂

[insu_na]

Whenever I create PRs to the projects I'm working on I try to have the PRs delete more code than they add (except when that's inappropriate ofc)
There's something satisfying making the codebase do something better, while decreasing complexity

[AlbertCloete]

"Engineers hate code. Especially code written by other people" most true thing I've ever heard.

[sasukesarutobi]

Upstreaming is definitely one way to help deal with Not Invented Here Syndrome, especially when it's a case of "this doesn't cover this one non-obscure use case we have."

[mbunkus]

Huge props for advocating for parallel & xargs. I use both daily, and the power they give me is immense. Not just efficiency/throughput (parallel processing), but also command chaining done right.

[elliott]

NIH syndrome is a huge problem, especially for smaller companies who really should focus more on their product than trying to invent some IP that is completely unrelated to their business. It's not just Sr. Engineers... sometimes it's CTOs as well.

[ShadowKestrel]

12:34 hit me on a way deeper level than it should have, lol. A few weeks back I was having some issues with xfwm, so I decided I would create my own desktop environment, entirely in Rust. A few days of coming to terms with my own skill issue later I looked into sway (i3 but wayland) and realised it does almost everything I wanted my DE to do, plus more, out of the box. Often the best way to do a task is to find someone else who already did it

[felipedidio]

Prime did the same thing to me that the Netflix guy did to him omg. I have changed my entire workflow because of it too. I used to run IntelliJ for Rust and avoid the command line as much as possible

[demolazer]

I hate it when code doesn't work. Then I find the solution after hours of frustration and then I feel like a god.

[STatic]

This is one of your best videos ever. The blog is fantastic and the commentary on top of it is "Uncle Bob Martin" worthy level of goodness here. I have already started blasting this to former co-workers.

VIM all the way!