"The End of Native Code"

[Daniel Keep]

David Medlock wrote:
> Andrei Khropov wrote:
>> David Medlock wrote:
>>
>>
>>> Walter Bright wrote:
>>>
>>>> pragma wrote:
>>>>
>>>>
>>>>> Slashdot had an interesting ask slashdot article yesterday about
>>>>> when  is
>>>>> it the right time to go whole-hog into interpreted/VM style language
>>>>> development.
>>>>>
>>>>> http://ask.slashdot.org/askslashdot/06/06/12/2044245.shtml
>>>>>
>>>>> As its an issue that we're all familar with, I figured I'd read
>>>>> and  see
>>>>> why this article had 1000+ comments.  I was actually quite suprised to
>>>>> find a  number of people, mostly C++ guys, clamoring for "native
>>>>> compilation plus garbage collection" or some variant thereof.  What
>>>>> was
>>>>> also suprising was the  number of "have you not seen D yet?"
>>>>> replies to
>>>>> these posts, and how well they  were modded *up*.  As slashdot is
>>>>> peer-moderated, this means that registered users  of the site had
>>>>> to take
>>>>> the time to hand out positive reviews on those  particular posts.
>>>>>
>>>>> So I have to say: you guys rock.  Remember, slashdot users pretty much
>>>>> hung D from the yard-arm on not one but two articles about D.  What
>>>>> I saw
>>>>> today was a subtle, but noticable shift in this attitude.  The word
>>>>> finally seems  to be getting out.
>>>>
>>>>
>>>> I saw the article when it first came out, but I obviously need to
>>>> go  read
>>>> the followups.
>>>>
>>>> The gist of the article as I interpreted it is that people go to script
>>>> languages because they are more productive. Why are they more
>>>> productive?
>>>>
>>>> 1) garbage collection
>>>>
>>>> 2) dynamic typing
>>>>
>>>> 3) lots of libraries
>>>>
>>>> D's got garbage collection.
>>>>
>>>> Dynamic typing is interesting in that while it is more productive,
>>>> it's  a
>>>> big reason why scripting languages will always be slooow. It's also
>>>> interesting in that if you look real hard at C++ templates, a lot
>>>> of  what
>>>> they are used for is to fake dynamic typing.
>>>>
>>>> D is moving towards what I call implicit typing - I've been taking
>>>> a  hard
>>>> look at where one is required to specify a type, and instead trying  to
>>>> figure out a way the type can be inferred instead (foreach is a good
>>>> example). Implicit typing gets D a number of the benefits of dynamic
>>>> typing with less complexity than the C++ template approach.
>>>
>>> <snip>
>>>
>>> Are you saying you will move towards SML like typing (Hindley-Milner
>>> type) ?
>>
>>
>> See Nemerle (http://nemerle.org/) - Hidney-Milner in C-family syntax.
>> Looks
>> very promising.
>>
>>
> 
> Thanks for the heads up.
> It does look good!  Looks like dot Net only though.. :(
> 
> I hope they get either native or even translation to C.
> 
> -DavidM

I've never understood this aversion to non-native languages.  Let's be
frank: native machine code is good, but it isn't the be-all and end-all.
 Nemerle was developed using .NET because that way they didn't have to
write a compiler; they could just use Reflection.Emit to generate their
assemblies.

Plus, it frees them from having to support every platform and processor
separately.

As for .NET itself, I think it's excellent.  For most purposes, it's got
plenty of performance (people are writing video decoders in C#), lots of
different languages to use, and a fairly complete set of libraries.

I agree in that it would be fantastic to have something like Nemerle for
native code, but I certainly won't discriminate against it just because
it's managed.

	-- Daniel

P.S.  Just in case you get the wrong idea, I'm far from a .NET fanb0i.
C# is only just "Ok" as a language, and the class library is painful in
many places.  Still, it's better than Java *shudders*.

-- 
Unlike Knuth, I have neither proven or tried the above; it may not even
make sense.

v2sw5+8Yhw5ln4+5pr6OFPma8u6+7Lw4Tm6+7l6+7D
i28a2Xs3MSr2e4/6+7t4TNSMb6HTOp5en5g6RAHCP  http://hackerkey.com/