Comment by maratd | Hacker News Reader

Comment by maratd | Hacker News Reader

Parent Full thread

> Why can't we have a standardized generic bytecode and VMs to run it? Why should the web be locked to a single language forever?

In the beginning, this was the intended design. Which is why you still specify the language when injecting code into the page.

That original design is flawed and does not work well on the web. Adding other languages is a horrible idea.

> Can you imagine if we had locked desktops and mobile only a single implementation language?

Apples and oranges. On both platforms, programs rarely interact with each other and when they do, it's through the OS using a standardized API.

The web is different. One piece of code is always interacting with another piece of code, that is interacting with a third and so on. There is no standardized API. You just call the appropriate function, if available.

To implement other languages, not only would you need to compile to bytecode, add VMs, etc. you would also need to add APIs that would allow communication between different pieces of code, written by different people, in different languages. Even if implemented, which would be a nightmare, it would still stifle the very nature of the web, the very dynamism that makes web apps different.

I don't want the web to become like the desktop or mobile space. I want to pull in code from twitter, facebook, my code, etc. and have it all work together, communicate with each other, and create a dynamic environment, all on one page, client-side. This is only coming into reality now.

The JVM and the CLR have already adequately demonstrated that multiple languages can easily (and with adequate performance) share common libraries and APIs.

Mixing Scala, Java, Ruby, JS, and PHP -- all in one process -- is simple and straight-forward on the JVM.

That said, the language of the web is HTTP and REST and open APIs, not injection/eval of JavaScript.

> The JVM and the CLR have already adequately demonstrated that multiple languages can easily (and with adequate performance) share common libraries and APIs.

Please tell me how code in PHP can get a response from a function in JavaScript? What if that response is a Date object? What if that Date object was extended with additional functionality? What if that Date object stored additional values in unexpected places? A huge advantage of Javascript is that everything is an object and everything can be modified.

When you try to make that work, you'll start placing restrictions on what can and what can't be passed. Global scope will be a nightmare. You'll try to hack together a scheme that casts responses into primitives. In the end, you'll end up with a stifling environment where interoperabilty is a theory that is never used in production because it was crippled to make it functional to the point of being useless.

Your concerns seem ill-founded given that these are solved problems. There are working implementations of low-barrier cross-language compatibility, today, for both the JVM and the CLR.

Here's one example; Rhino provides pretty seamless bridging from JavaScript to Java:

https://developer.mozilla.org/en/Scripting_Java

I've used it to run user-supplied JS inside the JVM.

PHP is pretty easy too:

http://quercus.caucho.com/quercus-3.1/doc/quercus.xtp#JavaPH...

Scala is even more lightweight here:

http://www.codecommit.com/blog/java/interop-between-java-and...

Bear in mind that these are all for the JVM, which wasn't even designed with the intention of being a general-purpose multi-language VM.

I can't speak to the JVM, but I know for a fact that some CLR languages have incompatibilities (C# and F# specifically).

> I've used it to run user-supplied JS inside the JVM.

Have you used Java libraries in JS or vice-versa? That's the problem the parent is referring to.

Both directions. There are impedance mismatches, of course, but it's nothing like the doomsday scenario he's claiming.

> Your concerns seem ill-founded given that these are solved problems.

Then solve the problem I put before you.

It's an easy one. I want to pass a modified Date object with stored procedures that modify the value based on geolocation. I want to pass it from a Javascript function to PHP/Java/whatever. Make it work. Please.

First of all, it's difficult to engage with your productively, as you're particularly confrontational.

That said, if you give me some example code, I'll demonstrate its use in a JavaScript/Java hybrid implementation when I have some spare moments.

This generally isn't the case with disparate languages on the JVM. Unless you have a concrete objection with a scientific basis, you're just spreading FUD, and I don't think that contributes to the conversation.

> Unless you have a concrete objection with a scientific basis

I did. Please solve the problem I put forward. If you solve it, you win.

You asked, "How can PHP get a response from a function in JavaScript?" Well, assuming we have the following JavaScript function:

  function multiplyString(str, times) {
    while (times) {
      str += str;
      times -= 1;
    }
    return str;
  }

We could write:

  <?php
    console->log(multiplyString("Hello", 3));
  ?>

You then ask, "What if the response is a Date object?" You would call it the same way, and get back a Date object.

The rest of your question adds a bunch of other suppositions, all of them fairly vague and none of which seem particularly more salient than the first two. If you have a particular reason to think this would be intractable, maybe you could explain in more detail the problem you see, and perhaps an example of where it shows up in popular multi-language runtimes and interfaces.

Reply on news.ycombinator.com