Hi,
there are a few problematic JIRA's related to the way those classes are
implemented. Mainly people are complaining on Vista and also when MINA
is used inside an applet, that it creates some ServerSocket and a Socket
in order to define the default socket/datagram configuration on the
system those applications are running on.
Despite the fact that this test is done in order to provide a set of
default values, it does more harm than it helps, as it's pretty
intrusive. We have to find a better way to do that, or may be remove
this feature from the framework.
Last, not least, it does not deal with ServerSocket options, which are a
sub-set of the Socket options (and I'm not sure MINA 2 covers all the
socket options anyway).
Here are the options supported by Socket/ServerSocket/Datagram on java 1.5 :
Datagram
--------
SO_BROADCAST
SO_RCVBUF
SO_REUSEADDR
SO_SNDBUF
SO_TIMEOUT
IP_TOS
Socket
------
SO_KEEPALIVE
SO_OOBINLINE
SO_RCVBUF
SO_REUSEADDR
SO_SNDBUF
SO_LINGER
SO_TIMEOUT
TCP_NODELAY
IP_TOS
ServerSocket
------------
SO_RCVBUF
SO_REUSEADDR
SO_TIMEOUT
As we need an existing socket to check this socket capability, the
current framework tries to create such a socket/datagram, and check for
the default options, setting some booleans accordingly. We can do it the
same way but on the first created socket, and stores the result in
static parameters. We just have to define a single boolean
(configInitialized) which will be turned to true as soon as we have
created the first socket/datagram/serverSocket. Of course, this part
must be synchronized.
Anyone has a better idea ?
--
--
cordialement, regards,
Emmanuel Lécharny
www.iktek.com
directory.apache.org
