Hi there,

Does anyone know a Microsoft email-adres for bug-reporting. I think I
found a bug in the StreamReader class:

Encoding dos = Encoding.GetEncoding(437);
StreamReader sr = new StreamReader(@"d:\test.in", dos);
StreamWriter sw = new StreamWriter(@"d:\test.out",false,dos);
string s = sr.ReadToEnd();
sw.WriteLine(s);
sw.Flush();
sw.Close();
sr.Close();

This should, in effect, duplicate the test.in file in the test.out
file. But, coincidentaly,
the test.in-file was a testfile of mine, containing 256 bytes 0xFF,
0xFE, 0xFD to 0x00.

And because 0xFF 0xFE is a unicode byte-order-mask (BOM), it thinks it
is an unicode file. This is obviously not the case, as I explicitly
provide a different encoding.

I think in this case, Microsoft should honor the encoding.

Just by using Lutz Roeder's excellent Reflector I found this bug; it
is actually in
System.IO.StreamReader.ReadBuffer():

[...]
       if (!this.IsPreamble())
       {
           if (this._detectEncoding && (this.byteLen >= 2))
           {
               this.DetectEncoding();
           }
           this.charLen += this.decoder.GetChars(this.byteBuffer, 0,
this.byteLen, this.charBuffer, this.charLen);
       }
[...]

The DetectEncoding() even fires when a encoding is supplied, which in
my opinion, shouldn't.

If anyone thinks I'm absolutely wrong, let me know:-)

Kind regards,

  Edwin.