This looks like a mcs bug. mcs don't generate a return statement at the
end of Main:
.method public static
default void Main() cil managed
{
// Method begins at RVA 0x20f4
.entrypoint
// Code size 25 (0x19)
.maxstack 8
IL_0000: br IL_0005
IL_0005: ldstr "Breaking..."
IL_000a: call void class [corlib]System.Console::WriteLine(string)
IL_000f: br IL_0019
IL_0014: br IL_0005
} // end of method default void Main()
csc generates correct code wit a return at the end:
.method public static hidebysig
default void Main() cil managed
{
// Method begins at RVA 0x2050
.entrypoint
// Code size 17 (0x11)
.maxstack 1
IL_0000: br.s IL_000e
IL_0002: ldstr "Breaking..."
IL_0007: call void class [corlib]System.Console::WriteLine(string)
IL_000c: br.s IL_0010
IL_000e: br.s IL_0002
IL_0010: ret
} // end of method default void Main()
csc also generates much shorter code!
- Dietmar
On Wed, 2002-07-03 at 08:29, Leigh Dyer wrote:
> Hey guys,
>
> I think I've found a minor bug in either mcs or the runtime. I was
> writing a small app that has just has a Main() and a small while(true)
> loop, with a break condition inside it. However, when I compiled my app
> with mcs and ran it with mono, after breaking out of the while(true)
> loop my program kept running - it seems to just keep re-running Main().
> Running with mint exits at the end of the first run, but with a
> "Trace/breakpoint trap" message.
>
> However, if I change my Main() so that it returns int instead of void,
> and add a "return 0;" to the end, everything works as expected.
>
> I tried to isolate the problem with a small test case, and while it
> doesn't exhibit exactly the same behaviour, it still doesn't work as
> expected with Main() returning void - mint segfaults and mono returns
> the following error:
>
> ** (process:18686): WARNING (recursed) **: unhandled exception
> System.NullReferenceException: "A null value was found where an object
> instance was required"
>
> RESULT: -1
>
> and then needs to be sent a ctrl-c.
>
> I'm running the mono CVS debs from http://www.atoker.com/mono/ as of
> yesterday on a Debian sid system.
>
> I've attached both my small test case and the original app (it's very
> small - all it does is iterate over the bytes of a text file looking for
> non-ascii chars) - I hope they're of some use.
>
> Thanks
> Leigh
> ----
>
> using System;
>
> public class BreakTest
> {
> public static void Main()
> //public static int Main()
> {
> while(true)
> {
> Console.WriteLine("Breaking...");
> break;
> }
> //return 0;
> }
> }
> ----
>
> using System;
> using System.IO;
>
> public class NonAscii
> {
> public static void Main(string[] args)
> //public static int Main(string[] args)
> {
> FileStream s = new FileStream(args[0], FileMode.Open, FileAccess.Read);
> int line=0, ch=0;
> int x=0;
>
> while(true)
> {
> int c = s.ReadByte();
> Console.WriteLine("char " + x + ": " + c);
> if(c == -1) break;
>
> if(c == '\n')
> {
> line++; ch=0;
> } else {
> if(c > 127)
> {
> Console.WriteLine("Non-ASCII char: line " +
>line + ", char " + ch);
> }
> ch++;
> }
> x++;
> }
> //return 0;
> }
> }
_______________________________________________
Mono-list maillist - [EMAIL PROTECTED]
http://lists.ximian.com/mailman/listinfo/mono-list
