maximum number of methods supported in C# class

[Sean Michael Murphy]

I let this run for an hour to get to 5000 before I gave up. Someone with more CPU and physical RAM than I have should run it and see where it ends...

using System;
using System.Collections.Generic;
using System.Text;
using System.CodeDom.Compiler;
   
namespace MethodCountLimitFinder {
   class Program {
      static void Main(string[] args) {
         Int32 methodCount = 1;
         Microsoft.CSharp.CSharpCodeProvider cscp = new Microsoft.CSharp.CSharpCodeProvider();
         ICodeCompiler icc = cscp.CreateCompiler();
   
         CompilerParameters cp = new CompilerParameters();
         cp.GenerateExecutable = false;
         cp.GenerateInMemory = true;
   
         CompilerResults cr = null;
         string pre = "using System;" + Environment.NewLine +
                  Environment.NewLine +
                  "namespace Tester {" + Environment.NewLine +
                  " class Test {" + Environment.NewLine;
         string post = " }" + Environment.NewLine +
                  "}";
         string inner = string.Empty;
   
         while (true) {
            inner += " public Int32 Method" + methodCount.ToString() + "() {" + Environment.NewLine +
                     " return 42;" + Environment.NewLine +
                     " }" + Envi

[DavidNohejl]

Sean Michael Murphy wrote:

while (true) { inner += " public Int32 Method" + methodCount.ToString() + "() {" + Environment.NewLine + " return 42;" + Environment.NewLine + " }" + Environment.NewLine; cr = icc.CompileAssemblyFromSource(cp, pre + inner + post); if (cr.Errors.Count > 0) break; methodCount++; if (methodCount % 10 == 0) System.Console.WriteLine(methodCount.ToString()); }

Sean Michael Murphy wrote:

Someone with more CPU and physical RAM than I have should run it and see where it ends...

No wonder, always use StringBuilder for string concatenation in a loop.

---

"Throughout human history, we have been dependent on machines to survive. Fate, it seems, is not without a sense of irony. " - Morpheus "Real men use mspaint for writing code and notepad for designing graphics." - Anna-Jayne Metcalfe

[Dan Neely]

Even assuming there is a constraint, you're talking about 2 totally different languages so why should they be the same?

-- You have to explain to them [VB coders] what you mean by "typed". their first response is likely to be something like, "Of course my code is typed. Do you think i magically project it onto the screen with the power of my mind?" --- John Simmons / outlaw programmer

[Martin 0]

Cannot been said to often! Good answere!

[PIEBALDconsult]

Well it can be said too often, but it's appropriate here.

[Sean Michael Murphy]

dnh wrote:

No wonder, always use StringBuilder for string concatenation in a loop.

Hmmm. Interesting. When I originally undertook to code this snippet to try to figure an answer to this guys question, optimization was pretty far from my mind. I mean, I cranked the original bit of code out in 15 minutes (or so) and had originally coded it so the methods would be recreated every time. I took another 5 minutes and optimized it so that only 1 method (the new one) would have to be concatenated to the "guts", which was then stuck in between the fixed "header" and "footer" of the class. It ran slowly, but I assumed that most of the overhead was in the actual code compilation (compiling classes of 15000 lines), and not a little bit of string concatenation. So I've re-written it using StringBuilder and timed both versions for 500 iterations. The original code did 500 iterations on my PC in 161.5222 seconds. This version:

StringBuilder inner = new StringBuilder();
 
DateTime startTime = DateTime.Now;
   
for (Int32 i = 0; i < 500; i++) {
   inner.Append(" public Int32 Method" + methodCount.ToString() + "() {" + Environment.NewLine +
                " return 42;" + Environment.NewLine +
                " }" + Environment.NewLine);
 
   StringBuilder code = new StringBuilder(pre);
   code.Append(inner);
   code.Append(post);
   cr = icc.CompileAssemblyFromSource(cp, code.ToString());
 
   if (cr.Errors.Count > 0)
      break;
 
   methodCount++;
 
   if (methodCount % 10 == 0)
      System.Console.WriteLine(methodCount.ToString());
}
 
TimeSpan ts = DateTime.Now - startTime;
 
System.Console.WriteLine(ts.TotalSeconds);

did it in 160.111. Much less that 1% slower. Not a string concatenation to be found, except for the line joins. Anything to add? Thanks. Sean

[DavidNohejl]

Sean Michael Murphy wrote:

Anything to add?

I'd agree that most time takes compilation, but the thing about string concatenation with + is that it's -unlike compilation - completely unnecessary. And I don't think that using StringBuilder for concatenating strings in big loops is optimalization - I think it's something you should do without thinking. btw you're still allocating 7 or so strings in

inner.Append("      public Int32 Method" + methodCount.ToString() + "() {" + Environment.NewLine +
                "         return 42;" + Environment.NewLine +
                "      }" + Environment.NewLine);

every cycle, that's 3500 unnecessary allocations :) Anyway, cool way to check for number of methods limit indeed.

---

"Throughout human history, we have been dependent on machines to survive. Fate, it seems, is not without a sense of irony. " - Morpheus "Real men use mspaint for writing code and notepad for designing graphics." - Anna-Jayne Metcalfe

[vytheese]

Hi, while( true) { do { inner.Append(" public Int32 Method" + methodCount.ToString() + "() {" + Environment.NewLine + " return 42;" + Environment.NewLine + " }" + Environment.NewLine); methodCount++; } while ((methodCount % 1000) != 0); cr = icc.CompileAssemblyFromSource(cp, pre + inner.ToString() + post); if (cr.Errors.Count > 0) break; System.Console.WriteLine(methodCount.ToString() + " Compiled successfuly ==> so not succed"); } System.Console.WriteLine(methodCount + " may be approximately to -1000 of method count"); I modified slightly your code as the above and executed, Its going on till 100000 ( above 1 lakh ), My machine got down, So I planned to run today night. Now I feeling, I shouldn't ask this question first of all ;) Thanks, Vythees -- modified at 5:24 Tuesday 3rd July, 2007 Thanks, Vythees

[Sean Michael Murphy]

vytheeswaran wrote:

Now I feeling, I shouldn't ask this question first of all

Don't be crazy. I enjoyed thinking about it. Sean

[PIEBALDconsult]

dnh wrote:

I think it's something you should do without thinking.

Never do anything without thinking.

[PIEBALDconsult]

A) I don't think there's any need for including the NewLines. B) Why step by one? Why not double methodCount after each successful compile?

[DavidNohejl]

I'll repeat: *Always* use string builder for concatenating strings in big loops. And I stay behind my claim. That being said, if that loop had about 5 iterations in 99,99% and much more in 0,01%, then you have to thinkg about it - IIRC StringBuilder would be slower. But if that task is something that must end in some very limited time or under very limited memory, you can't afford that 0,01% and even if performing worse in average, StringBuilder would be better choice.

---

"Throughout human history, we have been dependent on machines to survive. Fate, it seems, is not without a sense of irony. " - Morpheus "Real men use mspaint for writing code and notepad for designing graphics." - Anna-Jayne Metcalfe

[PIEBALDconsult]

In reference to A: error CS1034: Compiler limit exceeded: Line cannot exceed 16777214 characters The following succeeds at 1000000, and then I killed it while it tried 2000000.

namespace MethodCountLimitFinder
{
class Program
{
[System.STAThreadAttribute]
static void Main ( string [] args )
{
Microsoft.CSharp.CSharpCodeProvider provider =
new Microsoft.CSharp.CSharpCodeProvider() ;

        System.CodeDom.Compiler.CompilerParameters cp = 
            new System.CodeDom.Compiler.CompilerParameters() ;
        cp.GenerateExecutable = false ;
        cp.GenerateInMemory = true ;

        System.CodeDom.Compiler.CompilerResults cr = null ;

        System.Text.StringBuilder inner = 
            new System.Text.StringBuilder ( "namespace Tester { class Test {" ) ;

        int methodCount = 1000000 ;

        while ( true )
        {
            System.Console.WriteLine ( methodCount ) ;
            
            for ( int i = methodCount ; i > 0 ; i-- )
            {
                inner.AppendFormat ( "void M{0}(){{}}\\n" , methodCount++ ) ;
            }
            
            inner.Append ( "}}" ) ;
            
            cr = provider.CompileAssemblyFromSource ( cp , inner.ToString() ) ;
            
            if ( cr.Errors.Count > 0 )
            {
                break ;
            }
            
            inner.Remove ( inner.Length - 2 , 2 ) ;
        }

        foreach (  System.CodeDom.Compiler.CompilerError ce in cr.Errors )
        {
            System.Console.WriteLine ( ce.ToString() ) ;
        }
    }
}

}

-- modified at 21:11 Tuesday 3rd July, 2007 2000000 and counting...

C:\>maxi
1000000
2000000
4000000
error CS0001: Internal compiler error (0x80004005)
error CS0001: Internal compiler error (0xc0000017)
error CS0583: Internal Compiler Error (0xc0000005 at address 5A16E208): likely culprit is 'PARSE'.
error CS0586: Internal Compiler Error: stage 'PARSE'
error CS0587: Internal Compiler Error: stage 'PARSE'
error CS0587: Internal Compiler Error: stage 'BEGIN'

C:\>

-- modified at 1:56 Wednesday 4th July, 2007 After 3000000 I started hitting resource limits and timeouts. So now I simply have a program write a file with the code and compile it at the command lin

[Sean Michael Murphy]

PIEBALDconsult wrote:

A) I don't think there's any need for including the NewLines. B) Why step by one? Why not double methodCount after each successful compile?

Both excellent suggestions. 1) The NewLines was so I could preview the code during the initial stages of development. Same reason for the indents. I like even my autogenerated code to be neat and tidy. :) 2) Yup. Could have done a more efficient search, but was more interested in starting the app to get the result. By the time I had written the original and the slightly optimized version, I had spent 45 minutes and was getting tired of the exercise. And I thought that The Answer would actually be fairly low (thought it would probably be 256, 512 or 1024 max). I was surprised to see it climb over 2K, but kept expecting it to fail shortly. It never did, so I published the snippet and the result and encouraged others to continue in the work. The application was really intended as a starting point for figuring out the answer to this guys question. It was not a fully peer reviewed, optimized, documented, shrink-wrapped product, as you and others have adequately demonstrated by now... Sean