Thank you for your reply. I have just done some testing. Netscape
6.2 seems to refuse the reply messages. It FINs the connection immediately
after receiving the reply message. NS 6.2 can support SOCKS 5, but, I
believe, only with Sun JRE 1.4, which is still in Beta.
The other problems we are having are because our applet opens a
TCP connection back to the Web Server. Even though IE can deal with the
SOCKS Server, the Java VM apparently cannot. So with IE 5.0 and 6.0 with
Java VM 184.108.40.20667 and 220.127.116.1134, we get a "java.net.SocketException:
Malformed reply from SOCKS server" error. I don't know how to get IE
to use SOCKS 5, can you tell me?
>#Now still I'm not sure what does "the version of the reply code" mean,
>#and I think "should" should have been represented as "MUST" if other
>#values are treated as error:p
I completely agree that this specification must use the word
"MUST" here. The IETF has well defined meanings for these words, and I
was an Area Director a number of years ago, and I would not have let
this problem persist. But this is not an IETF specification, so we can
only argue about what the intent was.
The best advice I can give is what Jon Postel advised many years
ago: "Be liberal in what you accept, and conservative in what you send"
And by the way, do you know if any other SOCKS Servers also send
04 rather than 00 in their replies? For that matter, what other SOCKS
servers are generally in use?