Why doesn't C# have an 'inline' keyword?

re: Why doesn't C# have an 'inline' keyword?

Joeseph — Thu, 29 Jan 2004 22:52:00 GMT

Here's a related question.

When you create a property in .NET, we create it something like this.

private int _num;
public int TheNumber
{
get { return this._num; }
set { this._num = value; }
}

The compiler will create the appropriate get_ and set_ accessor functions.
My question is this.

If you are creating, say... a control, and using your properties internally, is it
more effecient to use the private member variable, or to call the property explicitly? Basically, should
you do ....
int temp = this.TheNumber
or....
int temp = this._num;

This is assuming that the getter function is just returning the value of the private member. If additional logic was needed, then of course you would call the accessor. But if you're just returning a simple value, I would think you could skip the overhead of calling the function. I doubt it would save much on 1 or 2 calls, but I've seen code similar to this many times.

for ( int i = 0; i != 1000; i++ )
{
if ( this.TheNumber > 8 )
// do something
}

In the above example, there would be a 1000 calls to the accessor function, which seems less effecient than 1000 calls to just check the value of a variable.

Encapsulation wise, it is far better to call the property directly, but is it better performance wise to call the private member directly?

I know there is no simple answer, but I'm curious as to whether or not the compiler helps out here.

re: Why doesn't C# have an 'inline' keyword?

Jason Salas — Thu, 29 Jan 2004 22:56:00 GMT

My desires are a bit more simplistic - I've just always wished C# would have something similar to VB.NET's "With" keyword. :)

re: Why doesn't C# have an 'inline' keyword?

W Poust — Thu, 29 Jan 2004 23:35:00 GMT

Several years ago our software group had the same discussion about inline functions. The senior developers made the same point that the compiler does a better job at optimizing. Actually between the compiler and the CPU instruction pipeline optimizations, its a non-trivial issue to understand. We told the gungho performance geeks that we wouldn't accept inline but if they were serious they could write asm blocks. Funny how none of the performance people took us up on the offer.

.NET method inlining

Jeff Key — Fri, 30 Jan 2004 04:20:00 GMT

re: Why doesn't C# have an 'inline' keyword?

Scott Mitchell — Fri, 30 Jan 2004 05:24:00 GMT

What would happen in VC++ with:

__forceinline void Factorial(int x)
{
if (x == 1)
return 1;
else
return x * Factorial(x-1); }
}

??? ;-)

re: Why doesn't C# have an 'inline' keyword?

Eric — Fri, 30 Jan 2004 05:57:00 GMT

Scott,

Human sacrifices, dogs and cats living together! Mass hysteria!

CLR Inlining

Paul's Imaginary Friend — Fri, 30 Jan 2004 12:09:00 GMT

Eric Gunnerson discusses Inlining

Code/Tea/Etc... — Mon, 02 Feb 2004 21:36:00 GMT

re: Why doesn't C# have an 'inline' keyword?

James Curran — Wed, 04 Feb 2004 17:31:00 GMT

Scott,
VC++ has an internal inline recursion limit (what? you think you were the first to think of trying that?). It default's to 8 (I think), but can be set higher with a #pragma.

Basically, the following code:
int a = Factorial(3);
int b = Factorial(7);
int c = Factorial(10);
can conpiled as if written:
int a = 6;
int b = 5040;
int c = 1814400 * Factorial(2);

And even in the out-of-line Factorial it has to create, it tries to inline some of the recursion. It's really fascinating to see the assembly code generated by that.

re: Why doesn't C# have an 'inline' keyword?

Adnan Masood — Thu, 05 Feb 2004 02:56:00 GMT

I appreciate Eric's memory with GhostBusters.

In reference to Microsoft Specific VC++ compiler, Scott's code would not be inlined since "its recursive and not accompanied by #pragma inline_recursion(on)"

C++ Language Reference
inline, __inline, __forceinline
http://msdn.microsoft.com/library/en-us/vclang/html/_pluslang_inline_specifier.asp?frame=true

Even with __forceinline, the compiler cannot inline code in all circumstances. The compiler cannot inline a function if:

The function or its caller is compiled with /Ob0 (the default option for debug builds).
The function and the caller use different types of exception handling (C++ exception handling in one, structured exception handling in the other).
The function has a variable argument list.
The function uses inline assembly, unless compiled with /Og, /Ox, /O1, or /O2.
The function returns an unwindable object by value, when compiled with /GX, /EHs, or /EHa.
The function receives an unwindable copy-constructed object passed by value, when compiled with /GX, /EHs,, or /EHa.
The function is recursive and not accompanied by #pragma inline_recursion(on). With the pragma, recursive functions can be inlined to a default depth of eight calls. To change the inlining depth, use inline_depth pragma.
The function is virtual and is called virtually. Direct calls to virtual functions can be inlined.
The program takes the address of the function and the call is made via the pointer to the function. Direct calls to functions that have had their address taken can be inlined.
The function is also marked with the naked __declspec modifier.

If the compiler cannot inline a function declared with __forceinline, it generates a level 1 warning (4714).

re: Why doesn't C# have an 'inline' keyword?

moo — Thu, 05 Feb 2004 12:25:00 GMT

JIT Inlining, is this set in stone on the first JIT or can it rejit if it realises its better inlined based on usage during runtime?

If not would this be implemented in future? Obviously you dont want it to spend alot of time reJITting a method to inline or not inline based on profiling data if its oscilating about.

re: Why doesn't C# have an 'inline' keyword?

moo — Thu, 05 Feb 2004 12:26:00 GMT

The world MIGHT end and planets MAY collide.

# re: Why doesn't C# have an 'inline' keyword? 1/29/2004 9:24 PM Scott Mitchell
What would happen in VC++ with:

__forceinline void Factorial(int x)
{
if (x == 1)
return 1;
else
return x * Factorial(x-1); }
}

??? ;-)

re: Why doesn't C# have an 'inline' keyword?

Adnan Masood — Thu, 05 Feb 2004 22:03:00 GMT

Also, the code was faulty since factorial is defined for n >= 0 and for 0!=1 by definition. (could be calculated too.)

n! = n*(n-1)! for n > 1

Therefore it should have been...

__forceinline void Factorial(int x)
{
if (x < 1)
return 1;
else
return x * Factorial(x-1); }
}

re: Why doesn't C# have an 'inline' keyword?

John Schultz — Mon, 26 Apr 2004 09:08:00 GMT

Adnan, the original Factorial() author's real question was "What happens with __forceinline and recursive fcns?", not "Does this code implement factorial right?"

Since you needlessly picked on someone else's pseudo code snippet, I'll pick on yours:

Your fix is only partial -- you still quietly give an incorrect answer for negatives, you don't detect overflow and you return void!

Of course the best implementation of Factorial, as you noted, is to just have an array of the first 20 pre-computed, 64 bit answers and use the argument as the offset.

Eric Gunnerson discusses Inlining

Code/Tea/Etc. — Wed, 25 Aug 2004 13:15:00 GMT

MBA

MBA — Sat, 18 Dec 2004 18:52:00 GMT

Helpful For MBA Fans.

re: How many bad developers are there in the world?

Software/Technology Discussion — Thu, 04 Aug 2005 20:30:03 GMT

re: How many bad developers are there in the world?

Software/Technology Discussion — Fri, 05 Aug 2005 04:27:33 GMT

Re: Inline routines

Channel 9 — Mon, 02 Jun 2008 00:40:03 GMT

.Net inlines everything it can

C# Inline Methods and Optimization | #2782 - Agile software development and best practices for building Microsoft .NET applications.

Wed, 20 Aug 2008 01:11:58 GMT

PingBack from http://www.ademiller.com/blogs/tech/2008/08/c-inline-methods-and-optimization/

Eric Gunnerson s C Compendium Why doesn t C have an inline keyword | Green Tea Fat Burner

Mon, 08 Jun 2009 05:37:27 GMT

PingBack from http://greenteafatburner.info/story.php?id=3567