Yesterday, I shared some spicy takes. A few were particularly controversial—most notably, that I correct Gif the correct way (with a soft G)—but I also got a lot of emails asking me to elaborate on a few of them.
Today, I wanted to talk about how tabs are objectively better than spaces. This won’t take long.
Tabs let you define how big you want each indent to be, and spaces do not.
What environment are you using that has a hardcoded tab size? I haven't seen this since typewriters.
Some projects just use tabs as a compressed form of 8 spaces. But that is a sin. Use tab to mean "one indent level" and align with spaces if you need to. (the occasional ASCII art diagram)
I think running tabs -N (where N is you preferred tab size) in the terminal should work. This is what I use in my zshrc on desktop.
SourceHut
Yup, they seem to be pretty opinionated here. If you look at the source there is just an inlined style with a single rule pre { tab-size: 8 }. I guess that is what you get when you use opinionated tools. The user's browser isn't right, my preference is right!
“View page source” in the browser
On Firefox this uses my default tab size of 4. But I guess changing this default isn't user-friendly.
Because other people might have restricted environment which might not suit their preference is not a good reason to level it down IMO.
Also, I think 9 is the best size for indent (matter of preference), do you think I should switch to space so everyone can enjoy this wonderful view I have ?
Oh, I've done my fair share of C++ and Python as well. But you got to agree with me that when you are on your fourth indented "if case" it's time to step back and think about what you are trying to achieve. I mean it's probably going to work, but probably also very hard to maintain that type of code.
There a many ways to implement abstractions, but it’s highly dependent on the language in question. You could simply refactor each level of nesting into its own function, with all dependents provided as parameters instead of scoped variables. You could then flatMap to avoid a bunch of nested looping, favoring a linear approach that’s often easier to reason about. You could go all out and refactor all your conditional statements away, in favor of the Either monad. You’d then have a number of functions, each doing one thing (including no nesting), and a main function gluing it all together, linearly. That is a pattern you can always apply; there’s nothing controversial about it, and on a similar note there’s nothing particularly challenging about Gaussian elimination.
…except when they don't. Many common environments have a hardcoded tab size of 8, which is insanely big for using it for indentation.
What environment are you using that has a hardcoded tab size? I haven't seen this since typewriters.
Some projects just use tabs as a compressed form of 8 spaces. But that is a sin. Use tab to mean "one indent level" and align with spaces if you need to. (the occasional ASCII art diagram)
I think running
tabs -N(whereNis you preferred tab size) in the terminal should work. This is what I use in my zshrc on desktop.Yup, they seem to be pretty opinionated here. If you look at the source there is just an inlined style with a single rule
pre { tab-size: 8 }. I guess that is what you get when you use opinionated tools. The user's browser isn't right, my preference is right!On Firefox this uses my default tab size of 4. But I guess changing this default isn't user-friendly.
Because other people might have restricted environment which might not suit their preference is not a good reason to level it down IMO.
Also, I think 9 is the best size for indent (matter of preference), do you think I should switch to space so everyone can enjoy this wonderful view I have ?
As an embedded software developer that does linux kernel drivers I've come to love the tab size 8 indentation level.
I'm paraphrasing: "if your indentation level gets too deep, it's time to rethink/refactor your function."
And with tab 8 you'll notice it rather quick if your function does too much/unrelated stuff.
A function should be short and do one thing only, if possible. It also makes unit testing easier if that's a requirement.
When you're operating on such a low level of abstraction, it's no wonder you don't need deep nesting.
Oh, I've done my fair share of C++ and Python as well. But you got to agree with me that when you are on your fourth indented "if case" it's time to step back and think about what you are trying to achieve. I mean it's probably going to work, but probably also very hard to maintain that type of code.
How would you implement, for example, Gaussian elimination with at most 3 levels of nesting?
Abstraction.
The solution for all levels of nesting.
Be specific. Which exact part would you abstract away and how?
There a many ways to implement abstractions, but it’s highly dependent on the language in question. You could simply refactor each level of nesting into its own function, with all dependents provided as parameters instead of scoped variables. You could then flatMap to avoid a bunch of nested looping, favoring a linear approach that’s often easier to reason about. You could go all out and refactor all your conditional statements away, in favor of the Either monad. You’d then have a number of functions, each doing one thing (including no nesting), and a main function gluing it all together, linearly. That is a pattern you can always apply; there’s nothing controversial about it, and on a similar note there’s nothing particularly challenging about Gaussian elimination.