[shevy-java]

> This reminds me of what zed Shaw said, for some reason code written without an ide is better and he's not sure why.

I am not sure whether the statement is correct; I am not sure whether the statement is incorrect either. But I tested many editors and IDEs over the years.

IDEs can be useful, but they also hide abstractions. I noticed this with IntelliJ IDEA in particular; before I used it I was using my old, simple editor, and ruby as the glue for numerous actions. So when I want to compile something, I just do, say:

    run FooBar.java

And this can do many things for me, including generating a binary via GraalVM and taking care of options. "run" is an alias or name to run.rb, which in turn handles running anything on my computer. In the IDE, I would have to add some config options and finding them is annoying; and often I can't do things I do via the commandline. So when I went to use the IDE, I felt limited and crippled in what I could do. My whole computer is actually an IDE already - not as convenient as a good GUI, of course, but I have all the options I want or need, and I can change and improve on each of them. Ruby acts as generic glue towards everything else on Linux here. It's perhaps not as sophisticated as a good IDE, but I can think in terms of what I want to do, without having to adjust to an IDE. This was also one reason I abandoned vim - I no longer wanted to have my brain adjust to vim. I am too used to adjust the language to how I think; in ruby this is easily possible. (In Java not so much, but one kind of has to combine ruby with a faster language too, be it C, C++, Go, Rust ... or Java. Ruby could also be replaced, e. g. with Python, so I feel that discussion is very similar; they are in a similar niche of usage too.)

[bccdee]

Simpler tools are a forcing function for simplicity. If you don't have code search, you'll need to write code that is legible without searching. If you don't have auto-refactoring utils, you'll have to be stricter about information-hiding. And if you don't have AI, you might hesitate to commit to the first thing you think of. You might go back to the drawing board in search of a deeper, simpler abstraction and end up reducing the size of your codebase instead of increasing it.

Conveniences sometimes make things more complicated in the long run, and I worry that code agents (the ultimate convenience) will lead to a sort of ultimate carelessness that makes our jobs harder.

[dijksterhuis]

> Simpler tools are a forcing function for simplicity. If you don't have code search, you'll need to write code that is legible without searching.

i was working in a place that had a real tech debt laden system. it was an absolute horror show. an offshore dev, the “manager” guy and i were sitting in a zoom call and i was ranting about how over complicated and horrific the codebase was, using one component as a specific example.

the offshore dev proceeded to use the JetBrains Ctrl + B keybind (jump to usages/definitions) to try and walk through how it all worked — “it’s really simple!” he said.

after a while i got frustrated, and interrupted him to point out that he’d had to navigate across something like 4 different files, multiple different levels of class inheritance and i don’t know how many different methods on those classes just to explain one component of a system used by maybe 5 people.

i used nano for a lot of that job. it forced me to be smarter by doing things simpler.

[jimbokun]

I really like this approach. A good reminder that Ruby started out as a shell scripting language, as evidenced by many of the built in primitives useful for shell programming.

[SoftTalker]

When .NET first came out I started learning it by writing C# code in Notepad and using csc.exe to compile it. I've never really used Visual Studio because it always made me feel that I didn't understand what was happening (that said, I changed jobs and never did any really big .NET project work).