Pretty printing of stackoverflow stacktrace does not always work

[jkotas]

Repro

Compiler and run on .NET 10:

M(1);

void M(int x)
{
   Console.WriteLine(x);
   M(x+1);
   Console.WriteLine(x);
}

Actual result

Very long stacktrace with at Program.<<Main>$>g__M|0_0(Int32) repeated thousands of time

Expected result

The repeating part of the stacktrace is eliminated, like:

Stack overflow.
....
Repeated 24121 times:
--------------------------------
   at Program.<<Main>$>g__M|0_0(Int32)
--------------------------------
   at Program.<Main>$(System.String[])

[mangod9]

assume this is a regression in 10?

[janvorli]

No, this has nothing to do with .NET 10.
The pretty printing can only handle case when the repeated sequence of frames is at the top of the stack trace. But in this case, the top of the stack looks like this:

Stack overflow.
   at Interop+Kernel32.WideCharToMultiByte(UInt32, UInt32, Char*, Int32, Byte*, Int32, Byte*, BOOL*)
   at System.IO.StreamWriter.Flush(Boolean, Boolean)
   at System.IO.StreamWriter.Write(System.String)
   at System.IO.TextWriter.WriteLine(Int32)
   at System.IO.TextWriter+SyncTextWriter.WriteLine(Int32)
   at System.Console.WriteLine(Int32)
   at Program.<<Main>$>g__M|0_0(Int32)
   at Program.<<Main>$>g__M|0_0(Int32)
   at Program.<<Main>$>g__M|0_0(Int32)
   at Program.<<Main>$>g__M|0_0(Int32)
   at Program.<<Main>$>g__M|0_0(Int32)
   at Program.<<Main>$>g__M|0_0(Int32)

It was always like this.

I'll try to think about possible ways to improve it while still keeping reasonable time complexity.

[mangod9]

Ok, if not a regression we can move to 11 then.