Hacker Newsnew | past | comments | ask | show | jobs | submitlogin

"Tabs vs spaces" is often misunderstood (and falsely reported) as a problem of preference.

The real problem is that using spaces for indentation is an accessibility issue.

The solution is to use tabs for indentation, and spaces for alignment.



> The real problem is that using spaces for indentation is an accessibility issue.

Not this again

https://github.com/prettier/prettier/issues/7475#issuecommen...

> Blind programmers have no problems with it.


Not everyone with vision or vision processing issues is blind. Being able to configure custom tab stops is an easy way to control what level of indentation is useful and clear.


As with all other types of data: the right approach is for model and presentation to be separable concerns.

We're struggling with the wrong problem if we aren't asking why the editor can't treat blocks as entities that are displayed however we want.



  > why the editor can't treat blocks as entities
That might with with C, but won't work with e.g. Python.


Python has syntactic blocks as well, and editors totally can (and do) muck around with their formatting. I don't see any reason why they couldn't handle customizable indentation.


The primary accessibility argument is that spaces waste precious space on braille displays.


There are perfectly good tools for that which don't require text changes.


> The solution is to use tabs for indentation, and spaces for alignment.

It's the second step that drives me nuts. Why can't/doesn't alignment happen at first step? The inconsistencies of alignment using true tabs, in a monospace plain text environment, grosses me out and decrements my faith in the underlying systems. I appreciate editors w/ settings like `insert spaces when pressing tab` and `the number of spaces a tab is equal to`.


Why would you ever mix them?! Even when I started programming 15 years ago it was already accepted wisdom that that was a terrible idea.


This is one of those things that started out as good advice, and then got turned in to an oversimplified parody of the original argument.

If you have something like:

  var (
    x     = foo
    other = bar
  )
Then clearly you should use spaces to align those "="s, no matter if you use spaces or tabs for indentation. Similarly, "hanging indents" like:

  coolFun(arg1,
          arg2)
Can only be done correctly with spaces, and it doesn't really matter if you use tabs or spaces for indentation; this will still align correct no matter the tabstop size:

  ->coolFun(arg1,
  ->        arg2)

  --->coolFun(arg1,
  --->        arg2)

  ------->coolFun(arg1,
  ------->        arg2)
If you had used tabs, changing the tab size would make it align all wrong.

The problem is when you start doing stuff like:

  if (one) {
  ------>if (two)
             bar();
  ------>else
             xxx(); 
  }
And then, depending on the tabstop size, things can start looking very misaligned and confusing; this is why Python 3 forbids mixing tabs and spaces in the same function, because it's so easy to make something appear wrong and there are no braces to clarify things.

That's what people mean with "don't mix tabs and spaces", not "if you use tabs then the space shall never appear in the file under any condition".


> If you had used tabs, changing the tab size would make it align all wrong.

This is clearly false. You said this after giving a perfect example of how it's done with tabs.


So you are saying that using a tabstop of eight to align this:

  -------->coolFun(arg1,
  -------->------->arg2)
Would still look nicely aligned with a tabstop of 4? and 2? Clearly that will not look right.


To understand tabstops properly, you need to throw away the "They always expand to N spaces." idea. If you have access to a mechanical typewriter with tabstops, then go and look at how it works (with tabs being set by protruding pins on the carriage). Tabstops in terminals in the Unix and Linux worlds work this way.

https://superuser.com/questions/710019/why-there-are-11-tabs...


If you use it as indention then "tab expands to n spaces" is actually accurate though.


You're misquoting yourself.


I don't know what your point is then, but this is turning in to a very trite conversation.


> > If you had used tabs, changing the tab size would make it align all wrong.

> This is clearly false. You said this after giving a perfect example of how it's done with tabs.

I believe the post you're replying to meant "used more tabs". That is, if you used any tabs beyond the code indentation, they would get altered when you change the indentation size and ruin the alignment.


This was indeed the misunderstanding.

When I made my original comment I thought what I was replying to was an example of "tabs to indent and spaces to align" immediately followed by a statement that the only way to use tabs is to both indent and align.


The basic idea is that you use tabs to align "blocks" (ie, multiple lines at the same indent level), and then spaces from there to align line "elements")

Aligning lines in the same block

    ____var a; // indented with a tab
    ____var b; // indented with a tab
Aligning elements of a line with the previous line

    ____var a, // indented with a tab
    ____----b, // indented with a tab, then aligned using spaces
The idea being that tabs are used where it makes sense that you could change them for preference and not have things look wonky... and then spaces are used in situations where a specific number of characters is necessary.

Another alternative is elastic tabs, where tabs are used for both of those, but are converted to an indentation/alignment number of characters semantically. I like the idea, but I've yet to see a good implementation.

Personally I'm a fan of all spaces, but that's mostly because every company I've worked at has used that.


> The real problem is that using spaces for indentation is an accessibility issue.

The problem of leading spaces not changing their size can be solved through programming.

Those programmers who have such a problem and need leading space to be rubbery, if those programmers are worthy, will solve that problem instead of complaining that everyone should adapt to them.

The use of tabs is detrimental to code bases. Whenever tabs creep in, invariably people use different tab settings and make a dog's breakfast out of the code.

Code exhibits indented sub-structures that are internally indented, but not themselves not aligned to a tab stop, but aligned with something. One kind of example is something like:

  function_name_of_random_length(lambda (a, b) {
                                     if (a < b) {
                                         indented();
                                     }
                                 },
                                 arg2);


Sometimes it can make sense to align arguments but I strongly disagree with pushing over a multi-line lambda that way.

So I'm not convinced the problem you're describing should ever happen.


While that may be true, "spaces for alignment" is nigh unsupported by all editors I've seen. They insist on replacing 8 (or N) spaces with tab even if in an alignment region.

int foobar(int a,

___________int b) // should only ever have spaces (using _ here because HTML)

But good luck finding an editor that won't insert a tab when you use the tab key here (willing to be wrong).

The other issue I have is that diffs break alignment between undented code and anything with tabs because the tabs "eat" the leading space, but unindented lines are offset by one.


I'm curious what editors you've experienced that with.

I've had some editors that made tab-space conversion an option, but it was never mandatory.


When I'm working in Git, Go, or Linux code, Vim (and NeoVim) uses `expandtab` in conjunction with `softtabstop` and converts any `tabstop` spaces into a tab (with `noexpandtab`) regardless of whether or not it is "alignment".

I've not done lots of development with other editors, but Kate didn't do it and the few times I used Visual Studio, it also had similar behaviors (I'm assuming VSCode inherited that stuff).


With a .editorconfig[1] file with the correct configs you can get most code editors to insert spaces instead of tabs.

[1] https://editorconfig.org/


What configuration is necessary for editors to insert spaces in code like this:

https://lore.kernel.org/keyrings/20221109025019.1855-2-linux...

If HN mangles the link, it is the addition of the `key_create` function in this thread: https://lore.kernel.org/keyrings/20221109025019.1855-2-linux...


Rustfmt, for one, understands how to mix tabs and spaces on the same line. I think the complication is overstated.


All JetBrains editors will put spaces.


I used to tout "tabs for indentation, spaces for alignment" until I started working primarily woth lisps. Idiomatic lisp indention isn't compatible with regular tab-stops, so the only solution is to go with spaces (maybe losing accessability) or, preferably, to go elastic.

https://nickgravgaard.com/elastic-tabstops/




Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: