Hacker News new | comments | show | ask | jobs | submit login

The key point:

  "ASI [“Automatic Semicolon Insertion”] is an error correction
  procedure. If you start to code as if it were a universal
  significant-newline rule, you will get into trouble."—Brendan Eich

I think the other thing that goes unmentioned is that a single semi-colon on a line in a file would pretty much end this debate.

Perhaps I've worked too much in teams, but surely "path of least resistance" has to factor some, right? I mean, if the choice is add a semi-colon to a line in a file versus asking somebody else to rewrite the compiler to be able to accept it, surely common sense would just be to add the semi-colon.

It doesn't mean that you're wrong, or that your code is broken, or that Crockford is smarter than you... but now you have code that works and compiles and that people can use.

Surely, that's gotta count for something, right?

Even if we inserted the semicolon, JSMin would still be broken :)


Sure it is the easier solution right now. But in the long run that would mean the Crockford would be able to dictate the style of Javascript to whatever the hell he wants.

Given how important Javascript is now (and will be in the future) we cannot allow any one person to retard the entire future.

Anyway use Googles closure compiler -- it can actually handle Javascript.

I think that's a visceral reaction based on his rudeness.

So long as he keeps building tools and utilities that people rely on, and that people use, he is at least influencing the direction of Javascript and, knowing that, I'd think one might take heed when Crockford says "this will break in the future", as he happens to have more insight into the language than most.

I understand that he wasn't polite, and I certainly understand the desire for the 'asshole' to be proven wrong. And maybe he is. Of course, that doesn't make him wrong, and it doesn't mean that he's "retarding the future" either, necessarily.

But the fact that jsmin is perhaps the most widely deployed JS minification utility does matter, whether anyone chooses to accept it or not.

And if you really think that he is retarding the future, the correct answer probably isn't to bicker about it on the internet, but to create a better compiler (or work on getting a better compiler accepted) as de facto.

At the end of the day though, Fat is the guy who has to deal with all the tickets talking about 'This doesn't compile in jsmin', and I'd think his life would ultimately be better off if he just added a semi-colon.

Of course, that's my opinion only, so take it for what it's worth.

But of course, we won't break existing code like those two lines from Bootstrap in the future (not per the promises/concurrency strawman, anyway). This was a point my blog took pains to make.

Doug may have forgotten the [no LineTerminator here] restriction, or he may not want it to the left of infix-! for promises, but I am certain that the whole of TC39 will not agree to such a breaking change.

we don't need to work on a better compiler then jsmin, because we have three -- uglifyjs, closure compiler, and yui compressor. They are all better then jsmin by any metric you choose, and they also don't break on automatic semicolon insertion.

That doesn't sound right to me. Making s language's syntax more liberal is not a sane way to correct errors. If anything, JavaScript's ASI is an error creation mechanism.

"Here is your new apartment. You must use a key to enter it. If you forget your key, you can buzz the superintendent during business hours, and if the super is home, (s)he will open the door for you.”

If you try to take your key with you at all times, the super will save your bacon once in a blue moon when you forget your key. On the other hand, if you think of the door as only needing a key when you wish to have it opened on nights and week-ends, you are on your way to trouble.

Of course other buildings simply don't let you leave without your key, but here at the JS Arms we feel that would be way too complicated.

This is not a good analogy. More like the superintendant will let you into a different apartment, or let you into your apartment and then punch you in the face. You just don't know.

But you don't need analogies to understand why ASI is a bad idea. ASI means that when you leave out a semi-colon you get unexpected behavior instead of a syntax error. Failure is always better than the unknown. At least then there's a chance the bug will be found and fixed.

Trusting artificial intelligence to make up for human stupidity doesn't look like a very good idea to me.

ASI is supposed to work in exceptional scenarios where you would have ended up making a mistake but wouldn't want to be reminded of. And that is supposed to be in rare scenarios.

Now if you make exception the norm and expect tools to make up for bad practices then its not going to help.

And this is why I think Python's forced indentation is in some way bad. Because it makes the code from a bad and good programmer both look same. And merely forcing code indentation won't magically transmogrify a bad programmer to a good programmer. There are many things to good programming and indentation is just one of them. Worse it will make both's code look the same.

Forgiving or masking or making bad practices look good doesn't help on the longer run. It only encourages such behavior. I am sure bad programmers can slip in easily into these communities than else where, because they are difficult to flush out and their mistakes are often forgiven or made look good.

> This is not a good analogy

Proceeds to offer an even worse analogy. Good one :)

I then went on to say that the analogy is unnecessary and offered an interpretation unaided by analogies. Did you think I was not aware of this?

That's a great analogy.

The difference is that if you forget your key, you know that you forgot your key, and you have to actively call the superintendent to have them "save your bacon." With ASI, you might never know that you had forgotten your key.

I agree that it would be nice to have something like "-Wall" that could warn you that you're doing things the "wrong way" but, uh, isn't that what JSLint is?

That's a terrible analogy. What's the analog for receiving a syntax error instead when you "forget your keys"? The analogy gives you bizarre and wrong choices and tradeoffs.

As usual, analogies do more harm than good, especially when understanding a relatively simple technical issue.

You forget your keys, the super doesn’t let you in, you’re locked out. That’s the equivalent of a syntax error.

On another topic, this isn’t really a technical issue, it’s a people issue. Everyone understands what the JS interpreter’s behaviour is, what Bootstrap.js does, and why JSMin doesn’t minify it. What is being discussed here is what choices people make to please themselves and others as opposed to the compiler.

If @fat didn’t care about people, he’d use semicolons and JSMin would compile Bootstrap. If Crockford didn’t care about people, JSMin would compile Bootstrap just the way it is. When we’re talking about multiple ways of writing code that does the same thing, it’s almost entirely about people and not technical considerations.

Yes, he's acknowledged many times (including in this post) that it was a mistake. That was the intention, but it wasn't successful. It should have gone all in (no semicolons at all) or not have existed.

I think it was a reasonable decision at the time. The idea was that JavaScript was not going to be some language people wrote hundreds of thousands of lines of code in, and that the target audience would become frustrated if the language didn't "help" them. For example, if JavaScript truly were to always require semicolons at the end of lines then that means this would be a syntax error:

<a onClick = "someFunc()" > ...

> Error, missing semicolon after ")" on line 1

That could get pretty annoying pretty fast, and simple little event handlers like these were probably the main use of JavaScript for the first x years. It also matched the existing philosophy of HTML which was to really go out of its way to make sense of whatever was in the file.

Now of course, the constraints are different: people want to use JavaScript for huge codebases and serious projects, so it makes sense that these decisions are no longer appropriate.

> For example, if JavaScript truly were to always require semicolons at the end of lines then that means this would be a syntax error:

Depends whether the semicolon is defined as a terminator or as a separator.

This would be valid code with statement separator semicolons.

Couldn't the 'onClick' case have been solved by allowing either a statement or an expression to appear there?

I believe he refers to error correction as in it corrects errors in the code -- i.e. the code itself should have semicolons and is fixed through ASI.

Right. To avoid confusion, I updated the post so the sentence reads

"ASI is (formally speaking) a syntactic error correction procedure."

Given that these are all formal languages, though, isn't that equivalent to saying that ASI is, formally speaking, part of the language's syntax? It's all just parsers! Apart from the social conventions around them, of course.

No, there's an important distinction. ASI does not even kick in without a syntax error.

Yet the "expectation of ASI" or (I think more likely) "expectation of newline significance" makes people believe that they'll get a ; inserted by separating two things by one or more newlines.

Most languages do not specify error correction. HTML5 of course does; CSS too; among general programming languages it's much less common. The spec for ASI does not fit in the tried-and-true LR(1) formalism used by ECMA-262. Parsing is not all ad-hoc or equally well-formalized and proven.

In addition to ASI, ECMA-262 has to use lookahead restrictions and a bit of semantic checking to cope with what could be purely syntactic concerns (say, if it could use GLR instead of LR(1)).

I suppose I'm thinking of that as implementation details; from the perspective of languages, it appears ECMA-262 does specify a well-defined language. Any input string is either rejected (not in the language at all), or is mapped unambiguously to an abstract syntax tree. So from that perspective, any sequence of characters that gets you an abstract syntax tree is a program in the language! How precisely it gets mapped is "innards of the parser" if you view languages as just string->AST mappings.

The error-correction view seems to want to add a third category, strings that are in some sense "errors", but nonetheless get unambiguously mapped to an AST. Which is strange from a classical formal-languages view, because if a string gets mapped to an AST, it's in the language, and the procedure that mapped it constitutes the parser! That category seems more like "warnings" to me, i.e. you probably shouldn't do this, but it will produce a program if you do.

Yes, ECMA-262 completely specifies (modulo bugs) sentences that are accepted or rejected, and for those accepted, their meanings.

But that doesn't alter ASI's error-correction nature, which is not an "implementation detail" -- it's in the spec and all too observable.

You're right, it has the character of a warning system, like Dart's unsound "types". But if it had been noisy (consoles in the early days were costly), too many developers would have ignored the warnings, and users would have paid for the overhead.

Your concluding sentence is spot on, I agree people should use semicolons in JS. Relying on a Ruby-like coding style in the large is way too risky.

I'm curious, does ASI actually hurt the performance of JS parsing ? I mean, will there be any performance difference between a code that properly uses semicolons versus one that relies on ASI..

It's interesting because the 5.1 specification doesn't say it is an error correction mechanism, it states: "For convenience, however, such semicolons may be omitted from the source text in certain situations".

Guidelines | FAQ | Support | API | Security | Lists | Bookmarklet | DMCA | Apply to YC | Contact