Tabs are objectively better than spaces - gomakethings.com
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 t...
Tabs let you define how big you want each indent to be
…except when they don't. Many common environments have a hardcoded tab size of 8, which is insanely big for using it for indentation.
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 ?
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)
Github uses 8 as a default. It's configurable though.
What environment are you using that has a hardcoded tab size?
Microsoft Windows' Notepad. I have sometimes used that when on a public computer.
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.
This is the biggest problem with tabs. Too many tools don't let you adjust the size (or make it very difficult). This is the only reason I usually prefer spaces (only very slightly).
My dream solution is elastic tabstops and I've posted about it here before a few months ago. The problem with wanting elastic tabstops is that it seriously compounds the issue of "editors don't properly support it"
Another accessibility reason for tabs: when using a braille display, each space takes up one character cell, so indenting with four spaces eats up four cells. Indenting three times with four spaces each eats up 12 characters already. Tabs only take one character cell each, so three indents = three character cells used.
The fact that there (I assume?) isn't a braille oriented text editor that can handle space-based indentation in a smarter way is a bit depressing. Maybe the solution should be better tools based around accessibility rather than convincing everyone to switch to tabs, which is a project that will just never succeed.
So your fix is “convince all the people that want/need the better handling to use a specific editor?” - perhaps it’s a smaller number of people, but do you not see the irony there?
I honestly don’t care about tabs vs. spaces, but if there’s a low cost change in my setup that makes it easier on others, why not?
My spontaneous reaction is that making some sort of braille oriented setting for some or hopefully most editors used by people with braille displays (I have no idea if using a "normal" editor even makes sense if you're using a braille display) is the most pragmatic solution to their screens being taking up by spaces.
First of all, convincing everyone to use tabs is a monumental task. Convincing people with braille displays to use more convenient tools on the other hand seems pretty easy, why wouldn't you want to use more convenient tools?
Secondly, there is a large amount of code written with spaces today, so even if people switch with tabs in the future you might still want to be able to read legacy code.
Thirdly, I don't think that the choice of tabs vs. spaces is completely arbitrary because of alignment. Using tabs for indentation and space for alignment leads to a lot more micro management of whitespace compared to just using spaces. I would guess that alignment isn't very braille friendly anyway, but it does make the code more readable for other people. Having a good braille editor affordance might be closer to letting us have our cake and eat it too.
Of course, I don't know what this would look like exactly, and maybe there's some sort of obstacle that I'm overlooking, I do want to be clear that this is just of the top of my head as someone who has never used a braille display.
That reminds me of those times when back on reddit some dev showed up to present their new GUI library. Bragging about how they were better than Qt devs etc. (even though they didn't implement the hard parts, like working text fields or tables)
After some time a bunch of people had enough and started bullying those guys into submission about accessibility. After some time, every of those toolkits had support or at least plans for supporting screenreaders. Eventually, AccessKit became a thing.
Good times.
Arguing that it’s invincible to convince people at large to adopt tabs over spaces with good arguments is a ridiculous statement
I do actually think that it is very hard to convince basically every programmer of something, no matter how good arguments you have.
Also, without knowing much about the issue, it sounds a bit like the tooling for people using braille displays isn't very good and fixing that is maybe also worth advocating for, perhaps it's even a strategy for advocacy that is more effective?
Yeah. It is depressing.
I’ve always wanted an accessibility feature that uses haptic feedback to mimic braille patterns for reading purposes too.
In general a lot of creative stuff can be done if we focused on it even a tiny bit more.
Neither tabs or spaces are good. The correct way is to leave no whitespace in the code at all. It's unnecessary and adds to processing time.
Everyone should aim for 1LOC per commit
My program, written in the whitespace language, ruined.
CURSE YOU PERRY THE PLATYPUS!
Now next time I read anything about why any Python libraries are named what they are named, I'm going to hear Dr Doofenschmirtz voice. Thank you for that.
Code formatter and run auto format
Holy wars are fun, but this is the answer.
At least this way the holy war takes place in commits to change the formatting file.
Unless go, in which case the holy war is now everyone who hates go format rewriting the code in a different language.
Tabs for indent, spaces for alignment. This is the way, I can't believe people are still fighting that ?
Anything for indent (barely matters, as long as the editor forces it to stay consistent), and fuck alignment, just put things on a new line.
I used to think this way, at least when writing C++. But it's objectively harder to do and convince other people to follow, especially if they can't be bothered to change their environment to display tabs and spaces differently. It's a losing battle so now I just do spaces when working with other people
I've yet to find tooling that supports this. Clang format has a setting that looks like it does it, but actually does something else. If I have to press the spacebar a bunch of times each time I add an argument to a function, that's a pain, and it's a bigger pain to convince the people I'm working with that that pain's less bad than using spaces everywhere and having the IDE deal with it.
Until the people making editors and auto formatters acknowledge that the obvious most sensible whitespace style is even a thing, I'm forced to do something else and be really grumpy about it.
I understand your point of view. Personally I either copy the previous line and replace the arguments there, or insert X number of space using the repetition feature of my editor. It also has a feature that will align multiple cursors together with the "farthest" one using space, which is a killer feature for me! (See this presentation video @1:40).
It's hard to do this consistently (especially in a team) because people might (and statistically in a large enough project, will) use the tab key for alignment since it's faster than pressing space, or just be confused about what whitespace is tabs and what is space. Just using space everywhere is idiot proof and requires no work to micromanage. The only way to use tabs is to not align at all.
And this is why language servers and formatters are so critical.
I agree that it's hard, but not impossible. This usually boils down to how Nazi people are when merging code. In a corporate environment, nobody gives a damn so yeah you gotta use whatever you want because there are already different indentation systems within the same file anyway :)
But hey, you gotta live by the changes you want to see happen, so I personally put a lot of effort in formatting my code regardless.
Then you lose the benefit of tabs: you can't adjust the tab width without destroying alignment. So you end up with a confusing mix of characters for no benefit.
Mixing them is the worst option.
The opposite is true, though. If you use tabs for indentation and spaces for alignment, you can adjust the tab width without destroying alignment. That's the big benefit of the tabs-for-indentation-spaces-for-alignment mix.
You can't do that with only tab characters, you can't even align stuff with tabs because it has variable width.
You might not understand how to do it properly so here's the idea:
Tabs will let you reach the indentation level of the current block, then from here, you'll use spaces to align stuff property. Here's an example, where >••• are tabs (I'm exaggerating alignment for the sake of the example) :
>•••if (condition1 == true
>••• || condition2 != false)
>•••{
>•••>•••struct ident people[] = [
>•••>•••>•••{
>•••>•••>•••>•••.name = "bob",
>•••>•••>•••>•••.pubkey = "value1",
>•••>•••>•••},
>•••>•••>•••{
>•••>•••>•••>•••.name = "alice",
>•••>•••>•••>•••.pubkey = "value2",
>•••>•••>•••}
>•••>•••];
>•••>•••secureConnection(people[0].name, people[0].pubkey,
>•••>••• people[1].name, people[1].pubkey,
>•••>••• CRYPTO_ALGO_DEFAULT);
>•••}
As you can see, everything will stay correctly aligned as long as it's within the same block.
You're confusing using tabs for indentation and spaces for alignment with using tabs and spaces for indentation. This means each line starts with tabs. Next you optionally have spaces for alignment with previous lines. Then you have content (like code or comments). Because you never have a tab following a space the alignment is never destroyed by adjusting how wide a tabstop is.
Years ago there was no way to share IDE settings between developers.
You ended up with some developers choosing a tab width of 2 spaces, some choosing 4 spaces and as there was no linting enforcement some people using 2-4 spaces depending on their IDE settings.
This resulted in an unreadable mess as stuff was idented to all sorts of random levels.
It doesn't matter if you use tabs or spaces as long as only one type is consistently used within a project.
Spaces tends to win because inevitably there are times you need to use spaces and so its difficult to ensure a project only uses tabs for identation.
IDE's support converting tabs into spaces based on tab width and code formatting will ensure correct indentation. You can now have centralised IDE settings so everyone gets the same setup.
Honestly 99% of people don't care about formatting (they only care when consistency isn't enforced and code is hard to read), there is always one person who wants a 60 charracter line width or only tabs or double new lined parathensis. Who then sucks up huge amounts of the team time arguing their thing is a must while they code in emacs, unlike the rest of the team using an actual ide.
inevitably there are times you need to use spaces
When? You indent with tabs then add any spaces you want for precise alignment. When would you need to use spaces to indent?
Do not mix tabs and spaces.
Its impossible to automate checking that tabs were only used for indentation and spacing for precise alignment. So you then take on a burden of manually checking
You end up with the issue where someone didn't realise and space idented or anouther person used tabs for precise alignment and people forget to check the whitespace characters in review and it ends up going inconsistent and becoming a huge pile of technical debt to fix.
Use only one, you can automate enforcement and ensure the code renders consistency.
I've always wondered why some people tout "forcing a consistent appearance across environments" as a pro for spaces. That's a bad thing.
To be honest I'm surprised code format converters aren't ubiquitous. Let the repo have it's master format, enforced on commit. Then converters translate into each developer's preferred standard dialect on checkout and back again on commit.
The consistent appearance thing is probably more about how mixing tabs (for indentation) and spaces (for alignment, eg in multi-line function definitions of calls) looks like complete crap if you change the tab width.
Using only tabs for indentation and only spaces for alignment will never result in crap alignment when adjusting tabstops because the alignment does not use tabs.
This is using both tabs and spaces for alignment.
--->func foo(int i, --->---> int j);
Observe what adjusting the tabs does,
->func foo(int i, ->-> int j);
This uses only spaces for alignment,
--->func foo(int i, ---> int j);
When converted the alignment is maintained because the tabstops aren't used for alignment, only for indentation.
->func foo(int i, -> int j);
I think you have it backwards. If you use tabs for indentation and spaces for alignment it works great for any tab size.
It is when you use a tab just as a compressed representation of 8 spaces and use them for alignment as well that it goes to shit. (because you have made the sin of tab == 8 spaces instead of the correct tab = 1 indent level)
Yesterday, I shared some spicy takes. A few were particularly controversial—most notably, that I correct Gif the correct way (with a soft G)
And I stopped reading there.
I consider tabs for indentation a failed concept.
The idea is good, but it evidently failed. Most guidelines and newer Tools recommend or require or use spaces for indent. They have their reasons too.
The prevalence of spaces makes it hard to make a contrary argument for tabs. By now, I don't think it's worth even if it had reasonable advantages.
Editors/IDEs that parse syntax can adjust space indent too. A mixture for indent and alignment is not obvious for everyone (I always display whitespace in my editors and am deliberate and consistent, but many people and editor defaults won't be). Some defaults of four or eight space-width tab display is atrociously wasteful and inaccessible.
Spaces are a good enough baseline. It works well enough. And most importantly it works consistently. That's why it won in prevalence and use.
If I could only get everyone who works on the thing I work on to use a whitespace visualizer, it would be enough. We can fight about tabs or spaces after we get rid of all the unnecessary trailing ones.
The argument for having tabs adjust depending on your ide sounds better than it is in practice. Someone formatting code to look nice with width 4 will look horrendous for someone who uses width 8.
Spaces makes it uniform and captures the exact style the original dev intended
If you have your tab width set on 8, that is on you. You will also set your IDE to insert 8 spaces when you press TAB and I will cry when I have to give you a code review.
When I indent my code, I am indicating that I am in a nested block. I don't care if, on your screen, that indent is 2, 3, or 4 characters.
I think calling one way better than the other is flawed. The reason the title is saying that tabs are objectively better is because they are used in addition to where spaces are used elsewhere. You could make the same argument in favor spaces due to keeping things simpler.
The argument of having variable indent size for tabs so viewers can decide how big they are is imho legitimate but also not the goal as it's addressing something that teams generally agree on. There is max characters per line, brace placement, general code style and rules. Yes we can eject the indentation from the rules that are agreed on but once again simplicity over complexity has an equal say.
In the end it doesn't matter that much, a good programmer will be able to work in either setting, the Editor will do most of the work anyways.
With all that said, spaces all the way!
Tabs take less space (space as in space in storage, like free as in freedom) tbh.
My disk is almost full! Better switch to tabs for indentation!
Happened to me quite a few times, relatable situation.
Seems there's not much else going on
Interesting take. I prefer spaces because each piece of code that I see with tabs has an implicit tabsize you really need to have if you don't want the code to look ugly - especially if the person has been mixing tabs and spaces - and they usually do. Sometimes unadvertently.
When you remove all tabs at least everyone is on the same page.
To the actual problem raised by the article:
I have ADHD. Two spaces per indent makes it damn near impossible for me to scan code. My brain gets too distracted by the visual noise. Someone who’s visually impaired might bump their font size up really large, and need to scale up or down the amount of space per indent. Someone might just prefer it because…
I wonder if it could be possible to adjust the "indent number of spaces you see" in code editors. Code editors are able to figure out what are indents and what are not, so in theory it should be possible. Perhaps that would be an idea for a new feature?
each piece of code that I see with tabs has an implicit tabsize you really need to have if you don’t want the code to look ugly - especially if the person has been mixing tabs and spaces - and they usually do.
Well written code doesn't have an implicit tab size. You should be using a tab to mean "one indent" and if you need to align something an exact number of characters then use spaces.
This is the downside to tabs, they are easier to use correctly. With spaces if it looks right in your editor it probably looks half decent everywhere else. Tabs have a worse behaviour if they are misused, but if used well then every viewer can view and edit with their preferred indent size.
I wonder if it could be possible to adjust the “indent number of spaces you see” in code editors.
You could potentially do a good job with a full parser for the language in question to determine the indent level and separate indent from alignment. But I’d rather not rely on this when we have a perfectly simple and semantic character for indicating what is an indent and what is an alignment.
Maybe this could be a useful linter though. That way mistakes are caught but not every editor needs a perfect parser of every language.
The thing is - I have probably seen hundreds of projects that use tabs for indentation ... and I've never seen a single one without tab errors. And that ignoring e.g. the fact that tabs break diffs or who knows how many other things.
Using spaces doesn't automatically mean a lack of errors but it's clearly easy enough that it's commonly achieved. The most common argument against spaces seems to boil down to "my editor inserts hard tabs and I don't know how to configure it".
Yes. That's what tabs are for. You can choose the width of the tab. It can be small for people with small screens. It can be big for this guy or people with 600 inch ultrawides.
Tabs are objectively superior because they are exactly what everyone wants at all times, and the git commit history does not get polluted.
laughs in lisp
Nah, I'll keep on sticking with spaces or whatever the language's formatter uses. Ain't no way am I mixing tabs and spaces, will just stick with spaces.
I used to be a tabs guy, somepoint over time, especially when I realized some of the edge cases I have in formatting only remain consistent when using spaces, I switched.
Looks like you missed the point. This is about indentation, not formatting.
Use tabs to indent your lines, but if you want to align a parameter with the parenthesis on the line above or something like that, you add spaces after the indentation tabs.
That way if someone wants to they can configure their tab length to 20 spaces and the indentation will remain consistent and the code will remain aligned.
why don't we store code unformatted and have everybody's IDE display it with their preferred format applied? it would make everything easier and stop people bickering over pointless things.
Storing an AST would be interesting, but it'd require the IDE to support parsing each specific language, so you'd probably want something like an LSP but for just parsing to handle that.
Ah yes I understand now!
I thought it was a non-issue that tooling should take care of anyway until stackoverflow published this:
https://stackoverflow.blog/2017/06/15/developers-use-spaces-make-money-use-tabs/
Spaces all the way
Make sure to specify you only use spaces before asking for the raise! It works!
Original poster is right by all accounts, of course. Now, let's come up with exotic significant indentations.
function xyz(a, b):
| var x = 2
| if true:
| | do_something()
| else:
| | do_something_else()
| anyway()
Pro: Your editor no longer needs to implement indentation hints.
Con: Looks obstructive if not highlighted like an indentation hint.
Your turn.
Right, that is another item I wished more editors would have picked up. Besides - say - nested language modes.
Wrong.
Honestly, my editor (Neovim) just picks between tabs and spaces for me, so I just end up using whatever's already there. The only language where I'll explicitly use one is Haskell, just because spaces there allow me to keep everything nice and lined up.
I'm tinkering with a whitespace language and prefer using 1-space rather than 2. I don't really like the double character for 1 level. Is that weird?
Tabs are forbidden though I could use tabs with a (1-line) per-file code filter for the compiler to turn 1 tab into 2 spaces, and that might be easier when working with others (though I don't know how it would be seen, especially needing to change editor tab behavior).
That’s all well and good if everyone uses editors you can configure to a certain standard all the time. Then tabs all the way.
Unfortunately that’s not reality for everyone.
Everyone should use a better editor.
Tabs let you define how big you want each indent to be, and spaces do not.
Spaces can too: Simply use more or less of them, to taste.
I have ADHD. Two spaces per indent makes it damn near impossible for me to scan code.
Then use four, or six, or eight, or 20. Hell, most code I've seen uses four spaces per indent anyway.
[Re: braille]
Surely there's an editor out there that will automatically display indent spaces as a tab character. Or failing that it seems like it would be rather trivial create a program to convert n spaces to tabs, and vice versa.
Spaces do not allow the viewer of code to choose how wide the indents are, this is dictated by the developer.
Most IDEs allow users to customise how many spaces to display tab indents as. Doing so the other way around may cause issues with languages based on whitespaces such as python.
You are missing the point. Lots of code has multiple authors. There is offer no space indention that works for all authors. With tabs each author or reader can use the width that works for them.
automatically display indent spaces as a tab character
You can't really do this reliably. The problem is that spaces may be used in other places for alignment where the width shouldn't be dynamic. If you do a simple s/ /\t/g you will have funny results where code was aligned carefully using spaces. (The reverse does work though if you want to go from tabs to spaces, because tabs contain more information.)
You could potentially do a good job with a full parser for the language in question to determine the indent level and separate indent from alignment. But I'd rather not rely on this for no reason. Sometimes I don't have a full parser available for every language I want to edit.