Fake getTcpNoDelay for SSL sockets to work

[CCob]

I've come across one use case (in a browser running an applet) that the SSL socket implementation is checking the property getTcpNoDelay on the NetSocketUDT. At the moment this currently throws an RuntimeException that it's unsupported.

Caused by: java.lang.RuntimeException: feature not available
    at com.barchart.udt.net.NetSocketUDT.getTcpNoDelay(NetSocketUDT.java:168)
    at sun.security.ssl.BaseSSLSocketImpl.getTcpNoDelay(Unknown Source)
    at sun.security.ssl.SSLSocketImpl.writeRecordInternal(Unknown Source)
    at sun.security.ssl.SSLSocketImpl.writeRecord(Unknown Source)

We should either return a true or false. This is a snippet from the OpenJDK implementation of the SSL socket.

        if (holdRecord) {
            // If we were requested to delay the record due to possibility
            // of Nagle's being active when finally got to writing, and
            // it's actually not, we don't really need to delay it.
            if (getTcpNoDelay()) {
                holdRecord = false;
            } else {
                // We need to hold the record, so let's provide
                // a per-socket place to do it.
                if (heldRecordBuffer == null) {
                    // Likely only need 37 bytes.
                    heldRecordBuffer = new ByteArrayOutputStream(40);
                }
            }
        }

Since UDT doesn't have a Nagle algorithm or similar, my gut instinct would be to return false instead of a RuntimeException. At least this should satisfy SSL. Whats your feeling on this?

[CCob]

I found out why it's doing it in the applet vs my test environment. My browser environment is HotSpot 1.7, vs my IDE environment of 1.6. It seems they have changed SSL in 1.7 to look at the Nagle info on the Socket.

[carrot-garden]

1. "fake getTcpNoDelay" sounds good
2. while you are at it, you may want to fake all else which currently blows up with "feature not available"
3. consider using log.debug() instead of exception so at least we know what is going on with missing features

[CCob]

I've done a pull request which puts in some sensible values for getters, but throw UnsupportedOperationException on setters, since an application may expect specific behavior to happen with these, so at least they will get a heads up.