c++0x is (finally) finalized

[Cogman]

http://cplusplus-soup.com/2011/03/28/c0x-will-be-c-2011/

Wow, that took long enough.

I'm actually looking forward to the auto variable type. Looks like it could be pretty useful for the lazy (especially when using iterators).

That, and the thread library is pretty interesting to me.

 

[Crusty]

Any idea when Concepts are going to be included again?

 

[Kirby]

So did they basically incorporate a few Boost libraries?

Threading FTW too.

 

[Cogman]

So did they basically incorporate a few Boost libraries?

Threading FTW too.

Added a few libraries, some small changes to how the language behaves (auto, rvalue, ect), and a few niceties (such as proper handling of unicode characters).

http://en.wikipedia.org/wiki/C%2B+0x

Also has for loops that are more compact. they are akin to foreach loops in other languages.

 

[BoberFett]

http://cplusplus-soup.com/2011/03/28/c0x-will-be-c-2011/

Wow, that took long enough.

I'm actually looking forward to the auto variable type. Looks like it could be pretty useful for the lazy (especially when using iterators).

That, and the thread library is pretty interesting to me.

Ugh, the variant. While it can be used properly by good coders who don't make assumptions about data, I've had to fix enough shoddy code with bugs introduced by weakly typed variables for a lifetime.

 

[Markbnj]

Ugh, the variant. While it can be used properly by good coders who don't make assumptions about data, I've had to fix enough shoddy code with bugs introduced by weakly typed variables for a lifetime.

Boy do I ever agree with this statement.

I don't mind...

var myList = new List<something>();

I do mind...

var myThingy = ReturnSomethingCompletelyUnknown();

Thank the lord for intellisense.

 

[Mark R]

Ugh, the variant. While it can be used properly by good coders who don't make assumptions about data, I've had to fix enough shoddy code with bugs introduced by weakly typed variables for a lifetime.

No. Auto is not a variant!

Auto is a fixed strong-type - however, it avoids the programmer having to specify the type in those cases where the type can be determined by the compiler.

e.g.

string [] mystrings = GetListOfStrings();
auto name = mystrings[0];

name = 1.0f;

will be pre-processed to

string [] mystrings = GetListOfStrings();

string name = mystrings[0];
name = 1.0f;

while will fail to compile because a float cannot be assigned to a strongly-typed string.

 

[Cogman]

Boy do I ever agree with this statement.

I don't mind...

var myList = new List<something>();

I do mind...

var myThingy = ReturnSomethingCompletelyUnknown();

Thank the lord for intellisense.

What it avoids (which I think is a good thing) is instead of typing something like

SuperLongNamespace::SuperLongClassName::SuperLongClassType varname = SuperLongClassName::function();

It doesn't allow variable type reassigning. Still strongly typed, just now you don't have to know the exact type you are assigning.

 

[Markbnj]

What it avoids (which I think is a good thing) is instead of typing something like

SuperLongNamespace::SuperLongClassName::SuperLongClassType varname = SuperLongClassName::function();

It doesn't allow variable type reassigning. Still strongly typed, just now you don't have to know the exact type you are assigning.

Agreed. All I'm saying is that I like to be able to read a declaration and know the type of the thing being declared, without having to navigate to a method definition or hover for the signature to see what it returns.

 

[Cogman]

Agreed. All I'm saying is that I like to be able to read a declaration and know the type of the thing being declared, without having to navigate to a method definition or hover for the signature to see what it returns.

True. And I guess it could add some confusion to what is going on since you have to be somewhat familiar with the code to know what is going on with different variables.

I wonder, could you do this

auto function(auto vara, auto varb)
{
   return vara + varb;
}

Because, honestly, that would eliminate most of the uses of templates.

 

[BoberFett]

True. And I guess it could add some confusion to what is going on since you have to be somewhat familiar with the code to know what is going on with different variables.

I guess I don't see the benefit. It may save a few keystrokes, but I prefer explicit code. Ambiguity makes for maintenance nightmares.

Also, I could see it causing issues with derived classes that may look similar due to usage at a glance, but have different functionality. I suppose I just have PTSD brought on by years of fixing horrible code by lazy programmers.

 

[Absolution75]

True. And I guess it could add some confusion to what is going on since you have to be somewhat familiar with the code to know what is going on with different variables.

I wonder, could you do this

auto function(auto vara, auto varb)
{
   return vara + varb;
}

Because, honestly, that would eliminate most of the uses of templates.

Auto can't be a return type or a parameter type

 

[JasonCoder]

Doesn't most of this stuff come from interoperability stories? Variants and IDispatch based interfaces came about to support scripting (typeless) clients. If you're not supporting that, you never need to deal with it.

Sort of like why would you do COM if you owned all the endpoints of an app. Think masturbating with a cheese grater... slightly amusing but mostly just painful.

If it's not a variant, then it's dynamic is the way I read it.