* Note: Whether the underlying socket is an SSL socket * can not necessarily be determined via {@link #isSecure}. * Plain sockets may be considered secure, for example if they are * connected to a known host in the same network segment. * On the other hand, SSL sockets may be considered insecure, * for example depending on the chosen cipher suite. *
* * @return the underlying SSL session if available, * {@code null} otherwise */ @Override SSLSession getSSLSession(); /** * Opens this connection according to the given route. * * @param route the route along which to open. It will be opened to * the first proxy if present, or directly to the target. * @param context the context for opening this connection * @param params the parameters for opening this connection * * @throws IOException in case of a problem */ void open(HttpRoute route, HttpContext context, HttpParams params) throws IOException; /** * Indicates that a tunnel to the target has been established. * The route is the one previously passed to {@link #open open}. * Subsequently, {@link #layerProtocol layerProtocol} can be called * to layer the TLS/SSL protocol on top of the tunnelled connection. ** Note: In HttpClient 3, a call to the corresponding method * would automatically trigger the layering of the TLS/SSL protocol. * This is not the case anymore, you can establish a tunnel without * layering a new protocol over the connection. *
* * @param secure {@code true} if the tunnel should be considered * secure, {@code false} otherwise * @param params the parameters for tunnelling this connection * * @throws IOException in case of a problem */ void tunnelTarget(boolean secure, HttpParams params) throws IOException; /** * Indicates that a tunnel to an intermediate proxy has been established. * This is used exclusively for so-called proxy chains, where * a request has to pass through multiple proxies before reaching the * target. In that case, all proxies but the last need to be tunnelled * when establishing the connection. Tunnelling of the last proxy to the * target is optional and would be indicated via {@link #tunnelTarget}. * * @param next the proxy to which the tunnel was established. * This is not the proxy through which * the tunnel was established, but the new end point * of the tunnel. The tunnel does not yet * reach to the target, use {@link #tunnelTarget} * to indicate an end-to-end tunnel. * @param secure {@code true} if the connection should be * considered secure, {@code false} otherwise * @param params the parameters for tunnelling this connection * * @throws IOException in case of a problem */ void tunnelProxy(HttpHost next, boolean secure, HttpParams params) throws IOException; /** * Layers a new protocol on top of a {@link #tunnelTarget tunnelled} * connection. This is typically used to create a TLS/SSL connection * through a proxy. * The route is the one previously passed to {@link #open open}. * It is not guaranteed that the layered connection is * {@link #isSecure secure}. * * @param context the context for layering on top of this connection * @param params the parameters for layering on top of this connection * * @throws IOException in case of a problem */ void layerProtocol(HttpContext context, HttpParams params) throws IOException; /** * Marks this connection as being in a reusable communication state. * The checkpoints for reuseable communication states (in the absence * of pipelining) are before sending a request and after receiving * the response in its entirety. * The connection will automatically clear the checkpoint when * used for communication. A call to this method indicates that * the next checkpoint has been reached. ** A reusable communication state is necessary but not sufficient * for the connection to be reused. * A {@link #getRoute route} mismatch, the connection being closed, * or other circumstances might prevent reuse. *
*/ void markReusable(); /** * Marks this connection as not being in a reusable state. * This can be used immediately before releasing this connection * to prevent its reuse. Reasons for preventing reuse include * error conditions and the evaluation of a * {@link org.apache.http.ConnectionReuseStrategy reuse strategy}. ** Note: * It is not necessary to call here before writing to * or reading from this connection. Communication attempts will * automatically unmark the state as non-reusable. It can then * be switched back using {@link #markReusable markReusable}. *
*/ void unmarkReusable(); /** * Indicates whether this connection is in a reusable communication state. * See {@link #markReusable markReusable} and * {@link #unmarkReusable unmarkReusable} for details. * * @return {@code true} if this connection is marked as being in * a reusable communication state, * {@code false} otherwise */ boolean isMarkedReusable(); /** * Assigns a state object to this connection. Connection managers may make * use of the connection state when allocating persistent connections. * * @param state The state object */ void setState(Object state); /** * Returns the state object associated with this connection. * * @return The state object */ Object getState(); /** * Sets the duration that this connection can remain idle before it is * reused. The connection should not be used again if this time elapses. The * idle duration must be reset after each request sent over this connection. * The elapsed time starts counting when the connection is released, which * is typically after the headers (and any response body, if present) is * fully consumed. */ void setIdleDuration(long duration, TimeUnit unit); }