Why Dart is not the language of the future.

http://blogs.perl.org/users/rafael_garcia-suarez/2011/10/why-dart-is-not-the-language-of-the-future.html

by vital101 • 15 years ago

260 116 15 years ago

blogs.perl.org

dextorious3• 15 years ago

May I point to the irony on posting why a language is not "the language of the future" on perl.org?

[UPDATE: yes, please, down vote freely, because humor is dangerous].

dillon• 15 years ago

I personally don't find an irony. Perl may not be the most popular language ever, but it definitely has its place. I do find an irony that the 0.1 specification of Perl is worst than the 0.1 specification of Dart.

prakashk2• 15 years ago

But, keep in mind that Perl 0.1 wasn't intended to replace a Javascript-like language. It was created as a replacement for Unix shell, and tools like sed, awk etc.

chromatic• 15 years ago

I do find an irony that the 0.1 specification of Perl is worst than the 0.1 specification of Dart.

Why is it surprising that a language designed to occupy a niche between shell and C in 1987 had a spec faster and looser than a language designed from (let's say) 2009 through 2011, based on an existing, popular, and well-understood language with well-publicized flaws?

njharman• 15 years ago

> Perl may not be the most popular language ever, but it definitely has its place.

Sure, but is it? Does it have any chance at all of being "the language of the future"? Or, is it a very old, mature language that has reached a plateau or is in slow decline?

mthreat• 15 years ago

I also find it ironic that this link to perl.org took 15 seconds to respond.

buster• 15 years ago

Despite all your downvotes i thought the same when reading (perl.org) behind the headline ;)

swah• 15 years ago

Those points just mean Dart isn't revolutionary, which is pretty much a requirement for a language to replace Javascript.

dextorious3• 15 years ago

Actually what you say makes perfect sense.

Say Google unveiled a Scala, Haskell or even Go like language.

Sure, the hip crowd would be pleased. But would it ever gain traction with the web dev masses?

Hell, those languages have not that much traction even outside web development.

Now, if Google had unveiled something like Ruby or Python, that would have also pleased the HN crowd (maybe a little less that some extravagant functional language with crazy type tricks). And it could possible gain traction too. But that too would hardly be revolutionary.

mhd• 15 years ago

One also has to remember that Google itself intends to use this language a lot, so it's much more likely that they approximate something they already use. And they're doing a lot of Java, where Dart doesn't seem too exotic. In some other thread, Dart was compared to GWT, and I think that's getting pretty close. For teams that would've used Java/GWT before, doing it in Dart instead isn't a huge jump. Probably not even as huge as Scala, where at least you'd be able to remain within the secure embrace of the JVM.

(Not a huge jump for Android programmers, either. In light of the Oracle lawsuit avalanche, this might be interesting…)

MatthewPhillips• 15 years ago

What is it about Go that would prevent traction where Dart will not. Not having classes?

chc• 15 years ago

Scala is gaining a lot of acceptance even without the backing of a major sponsor like Google. I daresay that if Google were behind Scala, we wouldn't be having this conversation.

jinushaun2• 15 years ago

Syntactically, Dart looks a lot like ActionScript 3, which is basically Javascript with classes. ECMA decided not to adopt Adobe's AS extensions into ECMAScript 4 because they felt it was trying to turn Javascript into Java. They wanted to maintain the "feeling" of coding Javascript. (Basically the paradigm pioneered by jQuery)

mmatants• 15 years ago

And I am so glad they did. I remember feeling unsettled by JS's lack of formal classes... and then I was enlightened.

mdaniel2• 15 years ago

I often wonder what the disadvantage is of moving to an existing, established, understood and already tooled language (such as the ActionScript family) if one is seeking to migrate away from a legacy language such as JavaScript.

Now I am using ActionScript to make a point, since I have not yet programmed in it to be able to address how much Dart does or does not mimic the AS syntax.

Is ActionScript "type-optional" as with Dart?

allertonm• 15 years ago

While it's true to say "ECMA decided not to adopt", it omits the important information that most of the opposition came from Yahoo (AIUI Douglas Crockford was instrumental in this) and Microsoft, who might have their own motivations other than "preserving the feeling of coding javascript." And Google too, which is kind of a WTF given Dart.

cubicle67• 15 years ago

no, it means that dart includes almost as much wtfery as js does. Things like not being able to say if (a) because unless a is a boolean true, then it's considered false. So we need to say if (a==a), except that there's no guarantee == will return a boolean, so it's possible (a==a) may evaluate as false!

If dart wants to replace javascript it has to be better, and by a long shot. at the moment I'm just not seeing that it is

unwind• 15 years ago

Having a == a be able to generate false is quite common, it's for instance the case in C, C++, Python and Ruby (off the top of my head), when a is the floating-point entity NaN. Surely not all of these languages are ill-defined?

sambeau4• 15 years ago

This is not a bad article despite the stupid title.

Take-up is all that matters in the future; the quality of a programming language has rarely had any influence on take-up.

qntm• 15 years ago

So what does influence take-up? And does Dart have it?

sambeau4• 15 years ago

I dunno, something like:

  * The backing of the largest web company in the world.

  * Inclusion in the second-most popular browser in the world.

Yep. That aught to do it.

guard-of-terra2• 15 years ago

We matter in the future. If we want to use a sane language in our future, we should probably help promising ones to take up and prevent more lame from doing so; the article is doing the latter.

mrspeaker• 15 years ago

Very interesting article. After playing with the examples and looking at the docs for a bit, it really felt "old school" - but I assumed that that was just a facade to lul any corporate java types into a false sense of familiarity before busting out the modern goodness: much like JavaScript's own "Surprise! I was really SmallTalk all along!" magic trick.

munificent• 15 years ago

> "Surprise! I was really SmallTalk all along!"

Self, actually. Dart is closer to Smalltalk in Java's clothing.

ohyes4• 15 years ago

There are a lot of unwarranted conclusions drawn here. I think this article is FUD (and I don't really care for Dart).

> Dart programs may be statically checked. The static checker will report some violations of the type rules, but such violations do not abort compilation or preclude execution.

> In other words the "static checker" is a lint-type development aid, not a language feature.

My reading is that the language is optionally typed (similar to SBCL). The type checker will warn you when you have a type declaration that appears to be incorrect, and will optimized based on these declarations, but will not keep you from running code that violates a type-check. This is a remarkably good solution for dynamic languages.

> Here's a strange thing: the one and only true value is the boolean literal boolean true. Everything else is false. That means that code in if (1) { ... } will never execute, because 1 is a number, not a boolean, and there is no implicit conversion to boolean. You'll need to write if (1==1) instead.

This makes as much sense as 'everything' being true. You would write if (true) {...}, but I can't imagine why you would do that.

> There is no implicit type conversion between numeric, string or boolean types.

Ok. This is a good thing. Auto conversion is bad.

> The distinction between string and numbers allows to re-use the addition operator + both for addition and concatenation. However, without strong typing, this will almost certainly prove to be a bad idea. From the specs, it looks like "2" + 2 will be a concatenation, and 2 + "2" a run-time exception (in the absence of implicit conversion from string to number), but experience infirms this: string concatenation happens in both cases (although with a warning in the second one).

This seems to directly contradict the first point. Which is it? It should do a dynamic type check and generate a run time exception in both cases. I don't see anything to convince me that "2" + 2 will be a concatenation... in the spec.

>Thus, isolates are a heavyweight thread control model very much like Perl 5's ithreads. That means that they are good for data isolation, but heavy to use and hungry in memory, because spawning a new isolate will imply cloning all the objects and data structures of the running libraries.

There is absolutely nothing that implies this. There is a tiny section in the spec about isolates. They could easily be lightweight like Erlang processes.

alttag2• 15 years ago

  > You would write if (true) {...}, but I can't imagine why you would do that.

In haste to make your point, you overlook that Boolean expression are used in loops too. Combined with a "break;" I'm sure many here have used a "while (true)".

nickknw• 15 years ago

He was just responding directly to the article's points. The author lamented that now he would "need to write If (1==1) instead"

darklajid2• 15 years ago

> This makes as much sense as 'everything' being true. You would write if (true) {...}, but I can't imagine why you would do that.

I think his real point is that there's no 'truthy', no evaluation to true. Later in the article he says that

  if (a) a.foo();

isn't possible because of that. A better example would've been (a != null) instead of (1 == 1), imho.

> This seems to directly contradict the first point. Which is it? It should do a dynamic type check and generate a run time exception in both cases. I don't see anything to convince me that "2" + 2 will be a concatenation... in the spec.

Not in the spec itself, right. But the current documentation seems to imply exactly what he states. See the definition of '+' here: http://www.dartlang.org/docs/api/String.html#String::String

"String +(Object other) Converts other to a string and creates a new string by concatenating this string with the converted other. "

ohyes4• 15 years ago

> A better example would've been (a != null)

This is a good practice, even better would be checking the type or class.

> "String +(Object other) Converts other to a string and creates a new string by concatenating this string with the converted other. "

This is a bad idea, now I have a gripe with dart.

seasoup• 15 years ago

>> There is no implicit type conversion between numeric, string or boolean types.

>Ok. This is a good thing. Auto conversion is bad.

One of the things I really dislike about developing in Ruby is that I can't do

value = 5 "The value is " + 5

I get bitten by this at least 10 times a day while developing, usually either when dumping to the log, or writing into html. I can fix the problem really quickly now, but every time I get an error:

"r" + 5 "TypeError: can't convert Fixnum into String"

I think, yes, yes you can easily convert a Fixnum into a String.

"r" + 5.to_s

It just means I have to sprinkle .to_s everywhere in my code. Ridiculous!

msy• 15 years ago

Just use inline implicit conversion: "r #{5+3/2}"

andrewcooke• 15 years ago

please can you explain what

    value = 5 "The value is " + 5

is supposed to do? at the moment i can't work out why it would be frustrating to be unable to type something that looks like a syntax error... (i'm assuming it has some meaning in a language i don't know, which i guess would mean php).

edit: what does "say" mean in the reply below? print? evaluate? and does "later" mean in another statement? or replacing what was before?

phaedryx• 15 years ago

I see that as a feature. It looks like you're saying "I want to add numbers and letters together and Ruby complains."

If you really want that value in your string, why not use "The value is #{value}"

rat2• 15 years ago

    "r#{5}"

or in actual use something like

    "I have #{me.num_of_children}\n"

much better then Auto conversion In python

    "I have " + str(me.num_of_sons) + "and " + str(me.num_of_daughters) +"\n"

does get annoying

regularfry4• 15 years ago

> This makes as much sense as 'everything' being true. You would write if (true) {...}, but I can't imagine why you would do that.

As I understand it, it's not that you might write if(1){} literally. It's that you might write a method to return <something> on success or null on failure, and ordinarily that would be reusable in a conditional. In Dart, it's not - unless you're careful. This is not the sort of thing we have grown up thinking that we need to be careful about.

Combined with the author's example of a==a potentially evaluating to either false or true depending on the class in a really unexpected way, I'm just seeing potential for confusion and I can't see any immediate reason why they would have thought that this would be a good idea. Yes, the spec mentions JS's problem with an autoboxed false, but they've just moved that problem around, they haven't solved it.

ohyes4• 15 years ago

> It's that you might write a method to return <something> on success or null on failure, and ordinarily that would be reusable in a conditional.

Doing this is like writing a pun into your code. Check the type of it, or check if it is equal to null (or if it is not equal to null).

malkia• 15 years ago

An example when a==a is when a is NaN

pjscott• 15 years ago

Dart will not actually optimize based on type annotations; it will optimize based on watching what the actual internal types of objects are, and generating specialized code for the common cases. From an interview with Lars Bak:

http://gototoday.dk/2011/10/10/lars-bak-on-dart/

rayiner3• 15 years ago

This sounds exactly like Strongtalk's optional type system.

fanf2• 15 years ago

He is right about isolates: see https://groups.google.com/a/dartlang.org/group/misc/browse_t... However it is wrong to compare them to Perl threads, because a new Dart isolate has a freshly initialized static environment, and is not a clone of the isolate that spawned it. Dart is designed to make static initialisation fast, so spawning should not be too slow.

jriddycuz• 15 years ago

The worst of both worlds: Dart fails to provides the advantages of static languages, without compensating by the flexibility of dynamic languages.

This was my impression exactly when I first read about Dart. This is the language that Google wants to replace Javascript? Please. Javascript may have some strange design quirks (or outright flaws, depending on who you talk to), but at least it has the sense to have a meaningful boolean context. I understand that you can't try to revolutionize everything if you're targeting a wide audience, but this just looks like decaffeinated Java.

Nitramp• 15 years ago

Meaningful boolean context? I understand that the original author is a Perl programmer which probably means he has wildly different views of the world, but seriously?

"Meaningful boolean context"s have given us things like the abundant "if (!!foo)" in JavaScript. And then he goes on to criticize absence of auto-conversions, which are the other plague of dynamic programming languages (see "===" & friends).

I don't see how anyone could actually want that kind of litter in his programming language.

alttag2• 15 years ago

I think one of the reasons JS succeeded is precisely a major complaint of experienced developers: globals [1]. But this, (perhaps as with PHP) is what enabled novice developers —some of whom eventually became experienced developers—to learn the language easily.

We all started with some bad practices, and as we matured our code did too. But the loose, forgiving standards are precisely what enabled broad use in the first place.

When a language imposes more constraints on the programmer (it may be for their own good!), it limits adoption and growth.

1: From the summary of Javascript: The Good Parts, "... Douglas Crockford identifies the abundance of good ideas that make JavaScript an outstanding object-oriented programming language-ideas such as functions, loose typing, dynamic objects, and an expressive object literal notation. Unfortunately, these good ideas are mixed in with bad and downright awful ideas, like a programming model based on global variables."

politician• 15 years ago

I think you're on to something there. Doesn't processing also encourage the use of globals as a way to make it more accessible to artists who are just learning how to program?

j_baker• 15 years ago

I think there are very few who would say that globals shouldn't be allowed altogether. The problem is that JS relies a bit too much on globals all being crammed into one namespace.

I'm an advocate of the python model (although I should maybe call it the Modula-3 model): only allow module-level globals. You get globals if you need them, but no worries about code mangling your variables.

cwp• 15 years ago

Actually, Gilad Bracha, one of the designers of Dart does say exactly that. He's also working on another language that has no static (ie, global) state. See http://newspeaklanguage.org/

mmahemoff• 15 years ago

This is what I like about CoffeeScript too. It wraps the module in a closure, so globals are opt-in. By default, the variable will be scoped to the generated module, but you can simply attach it to the global namespace (window.foo=x in the case of a browser-side CS).

techiferous• 15 years ago

Not quite. Ruby also has easy globals, meaning that you can start out learning it by writing tiny scripts and not worrying about creating your own classes, modules, multiple files, etc.

The difference is that JavaScript has a design flaw in that if you forget to declare a variable before you use it, it becomes globally scoped. This causes local variables inside functions, for example, to become accidentally global leading to bugs.

ajanuary• 15 years ago

The comments under the "Feeble Typing" section make it clear the author has never seen or read Gilad talk about optional typing. For one example, read http://bracha.org/pluggable-types.pdf

sadfasdfads2• 15 years ago

The way I understand it, Dart is a higher-level abstraction of Javascript. People thought that Java was crap too when it came out and said C/C++ was still king. Look how that went.

Perl has been out fucking forever and still isn't in the browser, so stop whining. Javascript whipped your ass in this arena, Perl.

sambeau4• 15 years ago

"People thought that Java was crap too when it came out and said C/C++ was still king"

I was there in 1995 and I have no recollection of anyone saying that. People were generally excited by its possibilities. The worry was about a) embedding a VM into a browser back when memory was limited and b) java was controlled by one company and wasn't an open standard.

Javascript won the 'browser war' because a) it was already there so it didn't need a plugin b) it was 'good enough' for its time c) it had Java in the title d) it had the sense to become an ECMA standard.

Perl never had any plans to be in the browser it was always a Unix command-line tool.

What is interesting is that Java started as a language for making UIs on iPad-like devices, The first web-server was written in Objective C, Perl gained traction as the way CGI programs were written. Now, Objective-C is used for programming iPads, Java moved from UI to Server and displaced Perl for corporate web apps. Meanwhile Javascript has quietly solidified its role in the browser and may finally make the leap to being a major server tech.

sadfasdfads2• 15 years ago

I was there in 1995, too. People were excited about the possibilities in Java and applets, because that was the only purpose in the beginning as far as most were concerned. Then people started writing UI apps in Java and starting to write other standalone apps and it was deemed "slow" and people made fun of it. People seem to forget that these days, because it wasn't long before it took off and became popular.

"Perl never had any plans to be in the browser it was always a Unix command-line tool."

Perl is not (just) a "Unix command-line tool". It is an interpreted language. It was a preferred language for old school webapps in the mid-to-late 90s, and continued in use a lot through the 2000s. Now for the majority, it has been replaced, except for code that is still running on old servers.

So why are Perl programmers ranting against Dart? Javascript is somewhat of a sore spot, even though the average Perl programmer would not admit to it and may not even realize it. A client-heavy web application that doesn't try to do things quickly saving keystrokes is about as far from Perl as one can get. Dart is "Javascript part 2" to the Perl programmer, and subconsciousnessly eats away at the Perl programmers ego. Even though they can't explain why, they feel as they must destroy it.

pyre3• 15 years ago

  > if (1) { ... } will never execute, because 1 is a
  > number, not a boolean,

This hasn't stopped Ruby from being popular.

scotth• 15 years ago

I guess you mean the opposite hasn't prevented Ruby from being popular. All numeric values are truthy, including 0.

koenigdavidmj• 15 years ago

A lot of Ruby idiomatic methods will return either a value or nil, which is falsey, so things like:

  x = meep
  if x
    # do stuff
  end

will run as most might mean.

pyre3• 15 years ago

  > I guess you mean the opposite hasn't
  > prevented Ruby from being popular.

I was stating something more like "the fact that there is a hard split between numbers and booleans, and everything -- barring null -- by default evaluates to true." Since everything is a object instance in Ruby, asserting a variable is basically only evaluating whether or not it is null.

saucetenuto• 15 years ago

In Ruby it's the other way around: 1 is truthy and the block will execute. Everything in Ruby is truthy other than the null reference and Boolean false.

I can understand forbidding non-Boolean values in Boolean contexts - it's not something I'd want myself but I think I get why a reasonable person might. Dart's semantics, though, are really strange: they follow no precedent I'm aware of, and don't seem to offer anything useful in themselves. Does anyone else have some insight here?

spooneybarger• 15 years ago

dart's boolean context semantics are like taking smalltalk's strongly typed ones and putting them through something.. i don't know what they were thinking. the basic idea is sound ( the smalltalk part ) but what they then did around the 'edges' is just downright off.

enmanuelr• 15 years ago

Yes, please. Tell me how this language, before it even launches, is finished, or has been tested in the real world, is already doomed to fail.

Oh and it's posted on a Perl blog. Completely objective I'm sure.

jbooth4• 15 years ago

Is it just me, or is every dart criticism follow the pattern:

1) I have a hobby horse 2) Dart gets released 3) Flog hobby horse

Go had a bunch of similar criticisms, mostly for not being "innovative" enough and favoring a small, solid feature set.

nickik2• 15 years ago

how is having classes, interfaces, generics, optinal types, operator overloading and functiontypes a small, solid feature set? Scheme, self that are languages with small feature sets.

jbooth4• 15 years ago

I was talking about go, which doesn't have generics, optional typing or operator overloading, doesn't really have "classes" (arguable), and a fairly tight type system which covers structs, interfaces and functions in a minimal way.

I don't really have an informed opinion on dart yet, I'm just noting that 90% of the complaints I'm seeing consist of people with a hobby horse that predated the Dart announcement.

bad_user3• 15 years ago

I don't know why you speak in the past tense, but I consider Go still not being "innovative" enough and also broken by design. We'll talk in about 2 years from now when they'll finally add generics to the language, in a broken way nonetheless, as generics do have to be baked in right from the start to pull it off. And it badly needs generics.

But you see, Go got all this attention just because it is a language released by Google. And in the meantime version 2.0 of a real systems programming language that is innovative and kick-ass is largely going unnoticed.

LeafStorm• 15 years ago

> And in the meantime version 2.0 of a real systems programming language that is innovative and kick-ass is largely going unnoticed.

Which one?

jbooth4• 15 years ago

FWIW I actually agree about the generics, both that Go needs them and that bolting them on afterwards typically doesn't work as well as having them from the start. But in general I wouldn't call it "broken" and view being "not innovative enough" as a feature for a systems language.

uriel3• 15 years ago

> And it badly needs generics.

Everyone I know who has actually used Go strongly disagrees with this claim, they are very happy with Go's interface system and most of them don't notice the lack of generics at all.

Go is not like other languages, and 'features' are not interchangeable across languages.

> But you see, Go got all this attention just because it is a language released by Google.

Yea, that it was created by Ken Thompson and Rob Pike had absolutely nothing to do with it. /a

digamber_kamat2• 15 years ago

I will not use dart because it comes from google. Wont let them own everything about the web.

digamber_kamat2• 15 years ago

I will still not use dart!

aristidb• 15 years ago

> Which means that the expression (a==a) might be, in some pathological cases, false.

Most languages (including Perl) have at least one case where a==a is not true. (0.0/=0.0) == (0.0/0.0), or any NaN == NaN expression.

regularfry4• 15 years ago

That's less a language thing than it is an IEE754 float thing, as I'm sure you're aware. It's justified there because it's a genuine propagation of an "undefined" result. I'm not sure that is the sort of protocol user code should be able to opt into by accident.

bryze• 15 years ago

If we're going to be posting judgmentally titled articles, then why don't we post "why perl is the language of the past." I suspect the main reason languages like perl and FORTRAN are still used is because of legacy and tradition. But I acknowledge that is my opinion, not fact, so I wouldn't post such a title. Seriouly, we need to stop the pomposity. What about "My concerns about Dart 0.1", or "Early Criticism for Dart 0.1"?

dreamdu5t• 15 years ago

Dart provides and enforces a bunch of idioms that you don't need in JavaScript in the first place! Adding classes to a language that... doesn't need classes.

If anybody but Google released Dart HN wouldn't give two shits.

rayiner3• 15 years ago

Like 80% of these things are true for Scheme/Lisp/Dylan. Canonical true value? Lack of implicit type conversions? These are good design decisions, not warts.

samth• 15 years ago

In Scheme/Lisp/Dylan/Racket/Clojure/Your favorite lisp dialect everything except #f is true -- Dart is the other way around.

rayiner3• 15 years ago

Right, it's the opposite convention from Lisp. I'm trying to distinguish from say Perl, which doesn't have a canonical true or false value but identifies a set of things that "seem like they should be false" like 0 and "".

nickik2• 15 years ago

The question is not what they share the question is what dart lacks. The two most importend things are (a) dart has statments and expressions (b) no macros.

Then dart does not have a nice concept of true/false witch most lisps have.

dos1• 15 years ago

To me Dart feels a bit like the Embrace/Extend/Extinguish behavior of Microsoft. There's nothing I've seen so far that really makes me excited about Dart, and it kind of feels like Google is really just creating this new thing for the sake of being different. It seems like they want to unilaterally make decisions on the future of the web. We all know how well that works.

bascule• 15 years ago

This article conflates "dynamic typing" with "weak typing" all over the place. First he whines that type declarations aren't mandatory (at least within the scope of what a type inferencer can resolve), claims the runtime won't be able to determine type information (which V8) can already do...

...then he goes on to complain that there's no automatic type coercion! This is probably one of the worst misfeatures of JavaScript (and Perl and PHP), but after a short tirade about "weak typing", he's longing for automatic type coercion? Sorry dude, "3 dog night" + 2 should not equal 5

roopeshv• 15 years ago

I'll just leave them here. (http://www.google.com/search?q=chrome+why+we+don%27t+need+an...)

http://www.uxpassion.com/2008/09/google-chrome-nah-we-dont-n...

http://www.extremetech.com/computing/79620-who-cares-about-g...

http://floggingenglish.com/2008/09/03/google-chrome-just-bec...

Why Dart is not the language of the future.

Related Stories