Oracle sues Google over use of Java in Android

Post by loup-vaillant » Sun, Aug 15, 2010, 9:51 AM UTC

barrkel wrote: post #300

loup-vaillant wrote: post #243

Earlier quoted context omitted.

For the particular cases I saw, runtime extensibility never mattered. (Do you have one where it does?) And extending a third party sum type is not very hard: data LibType = Zero | One Int | Two Int Int data MyType = ZeroOneTwo LibType | Three Int Int Int (It of course implies that you use MyType instead of LibType in your code.) Finally, if you really want to extend a sum type, no work around allowed, you might want…

The code using LibType is code you don't control; you can't change it to use MyType instead. Indeed, you don't even have the source code to it. That's also why the extensibility happens at runtime.

   new_function ZeroOneTwo x = old_function x
   new_function Three a b c  = -- handling new case

Post by KirinDave » Sun, Aug 15, 2010, 4:56 PM UTC

binaryfinery wrote: post #327

KirinDave wrote: post #141

Earlier quoted context omitted.

Read. The. Claim. And no, the CLR is not totally different from the JVM.

You've not looked at them have you (the VMs)? The JVM is entirely a class-focused JVM. The CLR is a general purpose VM. I can access arbitrary memory in the CLR, but cannot on the JVM. I write as someone who has attempted to build a byte-code assembly file that would allow direct manipulation of binary arrays for the purpose of efficiently loading OpenGL vertex buffers that mix floats and ints. Cant be done in Java o…

Post by loewenskind » Mon, Aug 16, 2010, 7:34 AM UTC

lsc wrote: post #318

loewenskind wrote: post #261

Earlier quoted context omitted.

I never bought into the whole "do no evil" bit, but I sure got sick of dealing with the army of people who did.

It's interesting to watch google... they are actually handling themselves /much better/ than I would expect them to... but I think there are solid business reasons for this. Google makes money because it uses data most of us consider private to serve up ads for us. If enough people stop trusting google, they will no longer be able to do this. Consumer trust is integral to their business. and there will always be tens…

Post by loup-vaillant » Mon, Aug 16, 2010, 8:51 AM UTC

barrkel wrote: post #333

loup-vaillant wrote: post #332

Earlier quoted context omitted.

So What? Just do what you would have done in Java without inheritance: aggregate. For each function that matters, you can write a new one that handle the extra case: new_function ZeroOneTwo x = old_function x new_function Three a b c = -- handling new case

I'm not sure if you're being willfully obtuse or not. The functions that matter are functions you didn't write. You don't have the source code to them. You cannot rewrite them with an extra case because you can't rewrite all the calling sites, because those are baked into executable binaries. The world of shrink-wrapped closed-source software may be a foreign world to many functional advocates, but it's the world I l…

  class LibType                { Handles 3 cases       }
  class MyType extends LibType { Handles a fourth case }

  // Statically typed, Class based OO language
  class Foo {
    int   bar(int,);
    float baz(float, int);
    int x;
  }

  -- Haskell
  data Foo = Foo (Int -> Int)
                 (Float -> Int -> Int)
                 Int

  -- The same, with record syntax. (for easy access)
  data Foo = Foo { bar :: Int -> Int
                   baz :: Float -> Int -> Int
                   x   :: Int
                 }

Post by binaryfinery » Mon, Aug 16, 2010, 7:02 PM UTC

KirinDave wrote: post #335

binaryfinery wrote: post #330

Earlier quoted context omitted.

And in evidence for the "Pretty Sure They Aren't" statement, here's Miguel de Icaza shitting his pants (not): http://tirania.org/blog/archive/2010/Aug-13.html

Miguel de Icaza, patent lawyer and prescient predictor. And also: unbiased.