An Open Letter to JavaScript Leaders Regarding Semicolons

http://blog.izs.me/post/2353458699/an-open-letter-to-javascript-leaders-regarding

by tswicegood • 16 years ago

84 77 16 years ago

blog.izs.me

nene2• 16 years ago

For me the reason to always use semicolons is the same reason I always use ===. There are just too many rules to remember and I've got more important things to keep in my mind.

It's also good to have your JavaScript look like the JavaScript most other people write. I have practiced many different coding conventions in my life, and I don't really hate none of them, but the thing I do dislike is context-switching between different conventions in different projects. We already have tabs v/s spaces and several other possible differences. It would be nice to exclude the choice for semicolon placement from that list.

I do agree with the author in one point: you should understand how automatic semicolon insertion works. And then make up your mind about it. Don't just blindly follow anybodys advice.

mistermann• 16 years ago

Maybe an inappropriate question, but could you explain the significance of === to the unenlightened amongst us?

nene2• 16 years ago

Comparing with === does not perform type coercion, while == does and the rules for type coercion in JavaScript are even more funky than for semicolon insertion.

adolph• 16 years ago

The standard equality operators (== and !=) compare two operands without regard to their type. The strict equality operators (=== and !==) perform equality comparisons on operands of the same type. Use strict equality operators if the operands must be of a specific type as well as value or if the exact type of the operands is important. Otherwise, use the standard equality operators, which allow you to compare the identity of two operands even if they are not of the same type.

So: (1 == '1') is true;(1 === '1') is false.

https://developer.mozilla.org/en/JavaScript/Reference/Operat...

scott_s• 16 years ago

I think this is the most important point: if understanding when or when not to use semicolons requires a subtle understanding of the language grammar, then terminating every statement has a cognitive cost. This cost adds to coding friction.

I don't personally use JS, but I do use C/C++ and Python. I prefer Python's syntax because I find it clean. Most of my work is in semicolon-and-brace languages, so I'm clearly familiar with that style, too. But were I to code in JS, I'd always use semicolons, because that's easier than constantly asking myself, "Do I need a semicolon or not?"

As an analogy, I use my turn signal every time I make a turn in my car. Every time. Even if I'm in an empty parking lot. That way, I never have to decide, "Should I use my turn signal?" It's just automatic. It reduces the amount of thought I have to put into things that don't actually matter.

trickjarrett• 16 years ago

I honestly never thought that much about the use of semicolons vs not.

I'm not a hardcore JS programmer, I know enough to get paid for it, and I'm the best JS programmer of my current employer. But I'm far from a jedi master.

Isn't this just a matter of personal preference? A semi-colon is interchangable with \n as long as you use it following the same rules. Which I do because I treat it as a visual marker for the end of a statement.

The counter argument is why use it if the system already recognizes the \n in place there, for which I say to make it easier for other JS programmers to recognize and parse since the \n is used in other places where it isn't the end of a line.

I guess I don't really see this as a compelling argument being made here, or I've misunderstood what is being argued.

JonnieCache2• 16 years ago

Most of the controversy stems from the fact that Doug Crockford, widely regarded as the javascript expert, has been aggressively saying "Use semicolons at the end of every statement or you will be miserable" for years, particularly in his influential book "Javascript - the Good Parts." You can see a presentation where he reiterates this point, among others, here: http://www.youtube.com/watch?v=hQVTIJBZook

This blogger has a pretty compelling argument against this view.

selectnull• 16 years ago

Actually, that's not what Crockford is saying. From the same link: "Figure out where the semicolons go. Put 'em in the right place. You'll be very much better off." That is a quote from 13:02 into the video. There is a trancript on the video, so you can read it too.

I won't claim Doug has never said something like that, but I doubt it. I watched many of his videos that can be found and he is always evangelizing learning the language, instead of blindly cargo culting.

robin_reala• 16 years ago

The author doesn’t seem to work in a team where the code is going to be touched by multiple people, all with varying levels of competence. It’s easy to explain to your colleagues that JS should have semi-colons added or weirdness can happen. It’s less easy to say ”Here’s the ECMAScript spec, go read it“ when they’ve got ruby and HTML and CSS and SQL to write and simply don’t have the time to thoroughly master each and every part of the stack.

olavk2• 16 years ago

The point is that always adding semicolons does not solve the weirdness problem. Consider:

    return
         17

This returns undefined, which is counter-intuitive if you come from other c-style syntax languages. But if you say "always terminate statements with a semicolons, otherwise weirdness happens", the intuitive fix is:

    return
        17;

Which still doesn't solve the problem. You have to remember an additional rule: "Don't insert line breaks in the middle of a statement, if the line break could be interpreted as a statement terminator." Or to put it another way: If you are in doubt, don't use a line break.

The problem isn't really if statements should be terminated with a semicolon or not. The problem is to decide when it is safe to insert a line break when it is not intended to terminate the statement.

gcr2• 16 years ago

Whoever linebreaks return statements like that deserves what's coming to them.

But I get what you're saying. Simple, consistent rules are better for teamwork.

cancelbubble• 16 years ago

He used to work at Yahoo. That's where I met him and while I never worked on a team with him, he seemed like a nice guy. Definitely smart, though his blog/writing style may come across as arrogant to some. I would have loved to end up on a team with him as I'm sure I could have learned a lot from him.

IsaacSchlueter12• 16 years ago

Thanks :)

mrchess• 16 years ago

This blog post reminds me of individuals who will only drink high quality naturally purified twice-filtered water from the mountains of Switzerland.

epochwolf• 16 years ago

Which comes from Lake Michigan?

IsaacSchlueter12• 16 years ago

How so? I missed the part where I wrote about swiss water.

Mr. Chess, I both admire and fear the potency of your absurdism. I would compliment you further, but I am too afraid.

davidsiems• 16 years ago

Here's what Crockford has to say on the subject:

"One of the mistakes in the language is semicolon insertion. This is a technique for making semicolons optional as statement terminators. It is reasonable for IDEs and shell programs to do semicolon insertion. It is not reasonable for the language definition to require compilers to do it. Use semicolons."

http://javascript.crockford.com/survey.html

I'm inclined to agree. You already have to deal with enough complexity while you're programming, why add more? It's easier to follow the rule 'always use semicolons' than to try to selectively use them.

mcantelon• 16 years ago

The counter to that argument is "You already have to deal with enough visual noise while you're programming, why add more?". Semi-colon insertion rules aren't that complicated. To me, it's a matter of what style project participants are comfortable with.

rimantas• 16 years ago

Rules may not be complicated. Seeing where and how they should be applied in the sea of code, some you did not write—entirely different story.

artsrc• 16 years ago

I agree use semicolons. Or longer:

Because JavaScript allows semicolon insertion some syntax is parsed in a way which is surprising to those skilled with other curly brace languages. This can causes problems if you do or don't try to leave out semicolons.

The solution to avoiding unexpected behavior is to run JSLint. JSLint will insist that you use semicolons. So you will have to use semicolons.

TamDenholm• 16 years ago

I think the reasons a lot of people use semicolons is that it's just easier. Also if you come from a language that enforces them, it's natural just to use them. That's why I use them anyway.

compay• 16 years ago

We programmers have a tendency to focus our arguments about code almost entirely from the perspective of code creation rather than maintenance, and I think that's in part what this article does.

An experienced coder who knows how Javascript works is unlikely to create problematic code even without semicolons, but it's fairly easy to imagine how errors could be introduced as a large, complex program is maintained by multiple developers, some of them potentially less experienced than the author. Like conditional braces in C, Javascript's semicolons are an optional syntax that can be used to make code more foolproof, and therefore more maintainable - though perhaps at the expense of some programmers' aesthetic preferences.

Really, though, I find most arguments about syntax aesthetics annoying, especially when people get their passions so inflamed over presentation rather than meaning that they have to risk "insulting" people. The stakes are so small. And if what you believe is so important, wouldn't it be better to convince people rather than intentionally alienate them?

People usually think there's a 1 to 1 correlation between "I like this" and "this is objectively good" and then spend a lot of time trying to logically justify their tastes. But taste doesn't really work that way - otherwise how else can you explain the popularity of Britney Spears?

I have my own tastes about how to write code in the languages I use, but I recognize that those preferences are generally just that - matters of taste - and tend not to try forcing them on others.

peti2• 16 years ago

I agree, but I would like to quote the author : "This isn’t an article where I try to convince you to write your code like I write mine." He does not really try to force his style. I respect his approach, because even if the article seems to be about aesthetic preferences, I think that he is concerned by the lack of curiosity and education that may be shown by some programmers. I think what the author really says is : RTFM. None of the comments here focus on the conclusion of the article, which is the most important part:

"Cozy up with some hot chocolate and the ECMAScript spec some Saturday afternoon. Practice a little. Play around with some test programs. It’s a good time. Or don’t do that, if you don’t feel like it. It’s your life. You almost certainly have better things to do with it. Just please stop making authoritative claims like “terminate all lines with semicolons to be safe.” It’s not any safer, or more reliable."

mattmanser2• 16 years ago

Some programmers like to know everything about a language they are using, knowing all the edge cases, keywords no-one else uses, the exact execution order of tokens and weird quirks. They are masters of complete understanding, and very occasionally this mastery allows them to come up with a revolution in the way things are done. Let's call these the Scholars.

A lot of others don't though, the joy of programming for them is not mastery of the grammar but writing the story. They would rather use a combination of run it and see and diving into the documentation only when totally necessary. Do not mistake this for an amateur who is cargo culting their code, these programmers know exactly what their code does, they just learn the bits they need as they need them. Let's call these the Mavericks.

I struggled to find neutral words for both schools of thought, don't read too much into the names I picked.

Both types can be excellent hackers. And of course we have a spectrum in between Scholars and Mavericks.

I am a Maverick. I find the Scholars, with their memorization of all the rules, to be both awe inspiring and entirely tiresome at the same time. I don't enjoy sitting down with a cocoa and a spec because I'd prefer to be programming. There is a vast swathe of every spec and every library that is entirely useless for the task I have at hand, quite often for any task I would ever do. I see anything which forces me to learn that useless knowledge as a waste of time.

Use semi-colons for every statement is a simple and easily remembered rule, one which takes no effort compared to reading a spec that delivers me little or even no benefit. I will also tend to forget these rules because my mind just doesn't see it as particularly useful information, I don't use it enough. I don't have any motivation to force that knowledge in my brain through learning practices because to a Maverick, it's boring.

There is no lack of curiosity or education in the latter type of programmer. I read this type of article which Scholars so often write with exasperation. Mavericks just don't care about learning the intricacies of language grammar compared to doing something else in programming. That something else is just as intellectually demanding.

Please remember that the next time you come across a Maverick, we're just of different schools, not of different ability.

joe_the_user• 16 years ago

I have a problem with the way the author puts forward his position: "I am sorry that, instead of educating you, the leaders in this language community have given you lies and fear. That was shameful."

As folks have said here, there is a valid alternative position. He's arguing as if there's no valid alternative ... which is problematic.

code_duck2• 16 years ago

I do not like inconsistency. I never leave out curly braces. I never omit semi-colons. There is no reason to do so.

I dislike working on code written by people who feel the need to omit these symbols. I would prefer if JavaScript would, like PHP or C, require semicolons at the end of every statement.

parkan• 16 years ago

Be very, very careful when you're using PHP as an example of the proper way to do something, because you're probably at least partially wrong. Semicolons are optional when followed by a closing tag (and likewise, closing tags are optional when the last statement ends with a semicolon -- in this case, whitespace at the end of the file is valid, as well)

code_duck2• 16 years ago

Yes. There is one situation in which you may omit a semicolon in PHP.

This does not create a situation like the semicolon insertion in JavaScript, however, and I don't perceive as an ambiguity or inconsistency issue.

Are you saying this detail about PHP causes problems of some sort? I can't think of any. If JavaScript were to work like that, and only let you omit a semicolon before a closing script tag, that would be absolutely fine with me.

axod3• 16 years ago

Use semicolons, always. Then you don't need to learn all the rules and caveats and ambiguity.

reid• 16 years ago

Actually, if you consider the return example in the blog, you do need to learn some caveats and rules. That's the point.

alextgordon• 16 years ago

No because you still have to remember that rule whether you use semicolons everywhere or not.

axod3• 16 years ago

Anyone who would write return\n deserves all they get.

It's a fun example, but it would never happen in real life.

ryan-allen• 16 years ago

Just because the rule exists doesn't mean it's the right thing to do.

Adding oddities to a language that could be small and simple creates these ongoing controversies. What are PHP developers _still_ talking about? SQL injection. What are Rails developers _still_ talking about? How to use REST properly. What are JavaScript developers _still_ talking about? Semicolon's or not.

This all points to poor design decisions in my mind. And we're at a loss to change the language unless we do something like CoffeeScript.

tlrobinson• 16 years ago

While there may be nothing technically wrong with Isaac's style, and it may in fact be slightly "better" in some way, it's clearly not how the language was intended to be written or how the vast majority of people write it.

I would be ok with leaving the semicolons off, if you didn't have to do ridiculous things like prepend some lines with semicolons instead. It's easier for everyone just to put the damn semicolons in.

I couldn't care less what your pants policy is in your own home, as long as I never come over. But when you invite me over (i.e. open source a project) and you have a ridiculous pants policy don't expect me to come back.

IsaacSchlueter12• 16 years ago

I would be ok with leaving the semicolons off, if you didn't have to do ridiculous things like prepend some lines with semicolons instead. It's easier for everyone just to put the damn semicolons in.

It'd be nice if I had a linter to catch things like this. But really, it's not so hard to spot. I was bitten by missing semis often enough to start prepending [ and ( already, before I started just leaving them out entirely.

No one's going to cry if you don't contribute to npm. When I contribute to your project, I'll follow whatever style you like, and if I find it to onerous, I'll move on. Life is tradeoffs.

ScottBurson• 16 years ago

And so once again, Alan Perlis' famous quip is proved right: syntactic sugar causes cancer of the semicolon.

Seriously -- look at the complexity of these rules, and then tell me that Lisp's parentheses are worse.

ShardPhoenix• 16 years ago

Javascript isn't exactly the typical case of semi-colon usage here.

jashkenas2• 16 years ago

I think that issacs is missing the part where other people try to read his code. We put semicolons at the end of each statement, even though it's shorter to leave them out, because it's how JavaScript is written, and how others are comfortable reading it. We spell English correctly, even though it's full of meddlesome "ou"s and "gh"s, because that's the way English is spelled. In fact, issacs' proposed semicolon-less, comma-first style reminds me of Mark Twain's "Plan for the Improvement of English Spelling":

    For example, in Year 1 that useless letter c would be dropped to 
    be replased either by k or s, and likewise x would no longer be 
    part of the alphabet. The only kase in which c would be retained 
    would be the ch formation, which will be dealt with later.

    Year 2 might reform w spelling, so that which and one would take 
    the same konsonant, wile Year 3 might well abolish y replasing it 
    with i and Iear 4 might fiks the g/j anomali wonse and for all.

    Jenerally, then, the improvement would kontinue iear bai iear 
    with Iear 5 doing awai with useless double konsonants, and Iears 
    6-12 or so modifaiing vowlz and the rimeining voist and unvoist 
    konsonants.

    Bai Iear 15 or sou, it wud fainali bi posibl tu meik ius ov thi 
    ridandant letez c, y and x — bai now jast a memori in the maindz 
    ov ould doderez — tu riplais ch, sh, and th rispektivli.

    Fainali, xen, aafte sam 20 iers ov orxogrefkl riform, wi wud hev 
    a lojikl, kohirnt speling in ius xrewawt xe Ingliy-spiking werld.

woodall2• 16 years ago

Sorry, had to try:

    For example, in Year 1 that useless letter c would be dropped to 
    be replased either by k or s, and likewise x would no longer be 
    part of the alphabet. The only case in which c would be retained 
    would be the ch formation, which will be dealt with later.

    Year 2 might reform w spelling, so that which and one would take 
    the same consonant, while Year 3 might well abolish y replacing it 
    with i and Year 4 might fix the g/j anomaly wronse and for all.

    Generally, then, the improvement would continue year by year 
    with Year 5 doing away with useless double consonants, and Years 
    6-12 or so modifying vowels and the remaining voiced and unvoiced 
    consonants.

    By Year 15 or so, it would finally be possible to make ius of thi 
    redundant letters c, y and x — by now just a memory in the minds 
    of ould doderez — to replace ch, sh, and th respectively.

    Finally, then, after some 20 years of orxogrefcl reform, we would have 
    a logical, cohernt spelling in ius xrewawt the English-speaking world.

treeface• 16 years ago

ould doderez = old dodders

orxogrefcl = orthographical

ius xrewawt = use throughout

chc• 16 years ago

Oddly, I found that mostly readable, though there were a few implicit changes that he failed to spell out, so you need to infer those.

IsaacSchlueter12• 16 years ago

npm is a hobby project designed to bring me joy. As a side-effect, a lot of node users find it helpful. That's great, and I'm proud, but honestly, if everyone stopped using it, I'd keep hacking on it.

A little boy may grow up to be the doctor who cures cancer. But that's not why you have children.

jashkenas2• 16 years ago

Yessir -- I'll vigorously defend your right to code in a style that floats your boat, but that's not what your blog post is: it's an evangelization of semicolon-free, paren-first style (hereafter known as SchlueterScript). You're asking the JS thought leaders to stop slandering your chosen syntax, and this comment thread contains a hundred reasons why most folks would rather not. I can dig the groovy art project also known as npm, and at the same time also rather not come across more SchlueterScript in the wild. A bit hypocritical, sure, but also heartfelt.

marijn• 16 years ago

Since my http://eloquentjavascript.net recommends always terminating statements with semicolons, I guess I'm being directly addressed by this.

The suggestion that 'I don't understand semicolon insertion' doesn't really hit the mark. I wrote a JavaScript implementation.

Yet still I recommend everyone to use explicit semicolons. Going over the proposed style on this page, it all sounds pretty good, and semicolon insertion is starting to look okay. Until you realize that he wants you to start every line that starts with [(+/-,.* with a semicolon. This is needed to prevent semicolon insertion from failing you.

I'm sure one can get used to such a style, but it is completely 'out there', radically disconnected from JavaScript's C parentage, and, though that's a question of taste, ugly and irregular. This solution is worse than the problem, I think.

quag• 16 years ago

"Until you realize that he wants you to start every line that starts with [(+/-,.* with a semicolon."

Good point, that isn't immediately clear.

To help me understand, the article listed an example where a line started with a [, what are situations where a line is likely to start with (+/-,.* ?

sharkbrainguy2• 16 years ago

The immediate-function pattern starts with an open paren.

(function () {

}());

IsaacSchlueter12• 16 years ago

I don't mind you telling people to end lines with semicolons.

I'd mind it if you told people that omitting semicolons causes cancer. (Or bugs.)

-- EDIT:

I just looked up your book to the point where you make this recommendation. You don't threaten any catastrophes, which is good. I disagree with the "complex and weird" conclusion, but it's certainly a valid decision to leave them out of the book if you'd rather not go into it.

My opinion of you: Not Evil. <3

Roboprog3• 16 years ago

Finally, somebody else who hates the largely gratuitous semicolons.

Example: http://roboprogs.com/devel/2010.06.html

It's a shame they didn't just make it fully line oriented (like BASIC, xBASE, shell-script, Python, Ruby, ...), with the customary backslash or ampersand to indicate an incomplete line. (but they didn't) I guess then Netscape would have had to call it FooScript (or something silly like echhh-muh-script).

The worst thing I hate about the "add semicolons!" religion is that it teaches people not to spot statements that are going to be terminated by a newline. So then, I have to see all the extra noise on each line :-(

Don't people test the crap they write?!?

richtaur• 16 years ago

Hopelessly pedantic … use them (or don't), but either way don't waste too much time on this subject. Focus on things that matter more.

IsaacSchlueter12• 16 years ago

I completely agree. I really don't care how anyone writes their code. I care about self-described experts giving false information about the language to newcomers.

If there is pedantry, it is about the importance of either telling the truth or shutting up.

jhuni• 16 years ago

JSLINT will catch basically all of these problems for you automatically, with it you don't have to worry about improperly parsed code, semi-colon insertion, ==, for in, the stupid constructors like new Function and new Object, implied globals, or most of the crap in the language.

There are still some deep problems with JavaScript that no tool, not even JSLINT can fix, like the fact that the language uses + for both concatenation and addition, there is no namespaces, there is no way to create immutable objects, etc.

I recommend you use a good IDE (like Eclipse) and hook it up with JSLINT so that you can ensure that you don't ever accidentally create global variables or create statements that the parser doesn't understand.

http://www.rockstarapps.com/joomla-1.5.8/products/jslint-ecl...

An Open Letter to JavaScript Leaders Regarding Semicolons

Related Stories