As i get older i opt for 4 spaces over 2 for a simple reason: My eyes are not getting better and reading 2 spaces is actually harder even though i'm only a little short-sighted.
I use 4 because that's how Xcode set it up. I used to use 2, but when you're working in a joint code-base without a linter (almost anything on GitHub) 4 spaces is the standard.
The only language I sometimes use 2 spaces for is HTML, since you can get so damn deeply nested. But even then I usually use 4 because it's so much easier to see the structure.
8 in C if the lead says 8 because in C you want people to limit the cyclotymic complexity of their code. C is hard to read.
4 in C when I code for fun or the lead says 4.
In fact 4 is a pretty decent visual aid and not to big for my use of 80 car/lines 25lines max per functions/method that I respect religiously and loosely) since 1998. I think being able to see the whole code without scrolling on a VT is kind of preparing yourself for the worst case (debuging in prod with vi/emacs or notepad) where attention is the key point under stress. Prepare yourself for the worst, the worst may happen.
In python, well, I pray that the lead does not mix tabs and space. Because, what a beautiful idea to have made indistinguishable characters significant. That is one of my huge beef with python as well as the inability to control the scope of variables (hence preventing strictures).
So I have strong opinions, but the lead wins when the lead is a benevolent dictator that proved to be right very often (like linus, theo, guido...).
The one who lead the project to its success makes the rules and as long as the project is going a way I like, I follow the guidelines happily.
"Now, some people will claim that having 8-character indentations makes
the code move too far to the right, and makes it hard to read on a
80-character terminal screen. The answer to that is that if you need
more than 3 levels of indentation, you're screwed anyway, and should fix
your program." --Linus Torvalds
I prefer tabs for indentation and spaces for alignment, but sadly, everyone seems to hate this. Either way, I adapt to the norm/standard of the language/project in question, like a sensible developer.
If everyone hates it, you've probably made the correct choice (for solo projects, at least).
With spaces, you not only repeat yourself needlessly, but you also lose the semantic difference between indentation and token separation (which is why converting spaces to tabs always leaves you with a few incorrect sections, whereas tabs to spaces does not).
But, again, it's tech, so the least sensible answer is usually the most popular. I wonder if it just makes some people feel manlier to write cryptic-looking camelCase with a crapload of redundant space characters.
Did you read my last sentence? The whole problem is that, despite consciously knowing better, programmers have a perverse urge, in the depths of their psyches, to make everything seem as difficult as possible. I can't think of any other reason why we still use camelCase, for example.
I think most devs prefer spaces to tabs (yes, literally typing in redundant spaces, while also losing some semantic content) Maybe it's because you always get consistent spacing that way, but maybe - just maybe - it's because of a puritanical urge to make programming harder.
Everyone in technology is aware of how important it is to keep things clear and simple, but deep in the soul, a lot of us just want to be Neo, typing bizarre hieroglyphs that only we ourselves are clever enough to decipher.
> I can't think of any other reason why we still use camelCase, for example.
I know you said that on purpose ;)
In any case, I am also a proponent of tabs for indentation, spaces for alignment, and in fact only recently did I succumb to using spaces for Python projects, for the sake of consistency with the rest of the ecosystem. I was merely pointing out how "many people use it, therefore it's probably bad", just like "many people use it, therefore it must be good" are never good arguments.
Well, it's never black and white, so, strictly speaking, what I wrote was wrong. But I still like my exaggerated version. It sounds a lot more dramatic.
So clearly, there are people who favor 2 and people who prefer 4. Solution: Indent with tabs, configure your editor, and everyone is happy. I don't get why people still promote the use of spaces.
I generally use spaces (in Python) to make up the difference when aligning. If you switched the tab size a bunch of visually indented code would become a complete mess.
The rule when using tabs is: indent with tabs, align with spaces. Tab everything out to the proper indentation level, and then if you want to align something then you use spaces on top of that indentation.
The problem is that when some line inevitable has space-based indent (which is basically impossible to see), then indentation goes crazy.
Of course, the same thing can happen with spaces, but it generally easier to fix. You can have a "no tab anywhere" policy, and easily enforce it. You can't have a "no space anywhere", or even a "no spaces at the start of lines", for example:
while(1) {
if (a == b && b == c && d == e &&
f == g && h == i) {
....
We need to tab-indent the 'if', but do tab+4 spaces the 'f == g'. That's very hard to automatically check and enforce
I personally turn invisible characters on so mistakes like that are a bit harder to make.
Along that line of thinking, the code should probably be reviewed by at least one other person before being officially committed, so one would hope that would be the place to find and fix such errors. Personally I've always felt the benefit of letting everyone set their own indentation level far outweighs the off chance of someone misindenting code (and I've never seen it cause a problem so far, not that it means much).
I do this all the time, I've never had issues, I tab as far as I can and space the rest.
As I said I have no issues with this, but if you really wanted, a quick regex for this would be something like:
Not allowed: " +\t"
Not allowed: "\t +\t" (this is redundant, but you get the idea)
That's only a problem if you use column alignment, as in your example. But column alignment isn't necessary in coding. It's quite possible - and easy - to write perfectly readable and maintainable code using indentation only and no column alignment.
Defining my terms a bit: By "indentation" I mean whitespace at the beginning of a line without any dependency on character widths. Extra whitespace in the middle of a line is "column alignment", and so is whitespace at the beginning of a line when the number of spaces depends not only on an indentation level, but also on matching other text in the previous line.
In your example, the "if" line uses indentation only, but the line beginning with "f == g" uses a mix of indentation (to match the indentation of the "if" line) and column alignment (the extra spaces to line the "f" up with the "a" above).
One good way to think about it is to imagine (or actually test) what the code would look like in a proportional font. If it still looks properly formatted, you're probably just using indentation. If the formatting gets messed up (e.g. the "a" and "f" no longer line up), you're using column alignment.
Avoiding column alignment has many benefits. If you ban it from your code, so many formatting issues go away completely.
I've written about this on HN a few times, so I'll leave the details to one of my previous posts:
3 spaces seemed to be the norm for block-structured languages (Pascal, PL/I, Cybil, PLUS, etc.) back in the "card image" days before tabs appeared in C source files. I think it worked because chains of END would line up in a nice diagonal.
I got used to 2 spaces in C++ in a big shop overrun with Style Guide fundamentalists and really long API names, but it's just not enough to convey logical nesting at a distance.
I've seen code, can't remember where, that used 4 space indentation for loops and 2 spaces for conditionals, so that indentation depth more accurately reflected runtime cost.
> They found that indentation level had a statistically significant effect on comprehension, and optimal indentation fell somewhere between 2 and 4 spaces.
25 comments
[ 2.3 ms ] story [ 77.3 ms ] threadhttps://www.python.org/dev/peps/pep-0008/#indentation
4 in C when I code for fun or the lead says 4.
In fact 4 is a pretty decent visual aid and not to big for my use of 80 car/lines 25lines max per functions/method that I respect religiously and loosely) since 1998. I think being able to see the whole code without scrolling on a VT is kind of preparing yourself for the worst case (debuging in prod with vi/emacs or notepad) where attention is the key point under stress. Prepare yourself for the worst, the worst may happen.
In python, well, I pray that the lead does not mix tabs and space. Because, what a beautiful idea to have made indistinguishable characters significant. That is one of my huge beef with python as well as the inability to control the scope of variables (hence preventing strictures).
So I have strong opinions, but the lead wins when the lead is a benevolent dictator that proved to be right very often (like linus, theo, guido...).
The one who lead the project to its success makes the rules and as long as the project is going a way I like, I follow the guidelines happily.
With spaces, you not only repeat yourself needlessly, but you also lose the semantic difference between indentation and token separation (which is why converting spaces to tabs always leaves you with a few incorrect sections, whereas tabs to spaces does not).
But, again, it's tech, so the least sensible answer is usually the most popular. I wonder if it just makes some people feel manlier to write cryptic-looking camelCase with a crapload of redundant space characters.
That's why I'm a Visual Basic Rockstar Ninja.
I think most devs prefer spaces to tabs (yes, literally typing in redundant spaces, while also losing some semantic content) Maybe it's because you always get consistent spacing that way, but maybe - just maybe - it's because of a puritanical urge to make programming harder.
Everyone in technology is aware of how important it is to keep things clear and simple, but deep in the soul, a lot of us just want to be Neo, typing bizarre hieroglyphs that only we ourselves are clever enough to decipher.
I know you said that on purpose ;)
In any case, I am also a proponent of tabs for indentation, spaces for alignment, and in fact only recently did I succumb to using spaces for Python projects, for the sake of consistency with the rest of the ecosystem. I was merely pointing out how "many people use it, therefore it's probably bad", just like "many people use it, therefore it must be good" are never good arguments.
Of course, the same thing can happen with spaces, but it generally easier to fix. You can have a "no tab anywhere" policy, and easily enforce it. You can't have a "no space anywhere", or even a "no spaces at the start of lines", for example:
We need to tab-indent the 'if', but do tab+4 spaces the 'f == g'. That's very hard to automatically check and enforceAlong that line of thinking, the code should probably be reviewed by at least one other person before being officially committed, so one would hope that would be the place to find and fix such errors. Personally I've always felt the benefit of letting everyone set their own indentation level far outweighs the off chance of someone misindenting code (and I've never seen it cause a problem so far, not that it means much).
As I said I have no issues with this, but if you really wanted, a quick regex for this would be something like: Not allowed: " +\t" Not allowed: "\t +\t" (this is redundant, but you get the idea)
Defining my terms a bit: By "indentation" I mean whitespace at the beginning of a line without any dependency on character widths. Extra whitespace in the middle of a line is "column alignment", and so is whitespace at the beginning of a line when the number of spaces depends not only on an indentation level, but also on matching other text in the previous line.
In your example, the "if" line uses indentation only, but the line beginning with "f == g" uses a mix of indentation (to match the indentation of the "if" line) and column alignment (the extra spaces to line the "f" up with the "a" above).
One good way to think about it is to imagine (or actually test) what the code would look like in a proportional font. If it still looks properly formatted, you're probably just using indentation. If the formatting gets messed up (e.g. the "a" and "f" no longer line up), you're using column alignment.
Avoiding column alignment has many benefits. If you ban it from your code, so many formatting issues go away completely.
I've written about this on HN a few times, so I'll leave the details to one of my previous posts:
https://news.ycombinator.com/item?id=10206860
I got used to 2 spaces in C++ in a big shop overrun with Style Guide fundamentalists and really long API names, but it's just not enough to convey logical nesting at a distance.
I've seen code, can't remember where, that used 4 space indentation for loops and 2 spaces for conditionals, so that indentation depth more accurately reflected runtime cost.
There's your answer. 3.