fixes microsoft/vscode-mono-debug#104

[walcht]

No description provided.

[akoeplinger]

This doesn't sound correct to me, the protocol says the headers/content should be separated by carriage return, line feed and not the literal string backslash-r, backslash-n.

[walcht]

Yes, you are correct... Please ignore that commit.

Just to give some further context into why I opened this pull request, I described the issue at #104.
I tried to use NeoVim with nvim-dap client (not relevant since I am pretty confident the issue is not in NeoVim client side). I simply get a timeout to the initial initialize request.

After looking at this particular line (at: https://github.com/walcht/vscode-mono-debug/blob/main/src/csharp/Protocol.cs?plain=1#L197):

var idx = s.IndexOf(TWO_CRLF);

I realized that the issue was because of a change to IndexOf implementation for newer dotnet versions that completely breaks the TWO_CRLF detection. The exact issue is described in detail here: dotnet/runtime#43736

I managed to 'solve' this in a similar project by not using IndexOf() and instead finding the literal "\r\n\r\n" in the string.

CONTENT_LENGTH_MATCHER = new Regex(@"Content-Length: (\d+)\r\n\r\n");

Although this solved my issue, searching for a literal "\r\n\r\n" character sequence is not correct (as per the protocol).

Which brings me to this question:
I think I am missing something here - why when I use verbatim string interpretation, all issues are solved and everything works fine?

(Thank you for responding and putting time into this - I really appreciate this project.)