> How much faster it will be (in your opinion) ?
It may be faster. However, performance is not the
reason for the change.
> Is this will be working like
> serialization/unserialization Java objects into
> XML strings ?
Not quite. It will change the way how the argument
list is build and how the client receives server
replies. The three server requests (50..53) will
remain, all other sub-requests are replaced by sending
a text-based description of the arguments:
<C i=ID>
<A i=ID v="Hello World"/>
<O i=ID v="0xAF6D"/>
</C>
These arg lists can be parsed efficiently and we avoid
jni calls/round-trips since we don't have to ask the
server for the method ID's, global refs etc. to build
a binary argument list.
> Again, if you will need testing just write...
Yes, thank you. My plan is to check in the new
protocol code this weekend. After that we can test
the implementation. Since the protocol is text based,
debugging it should be much easier than before.
Jost
___________________________________________________________
Gesendet von Yahoo! Mail - Jetzt mit 250MB Speicher kostenlos - Hier anmelden: http://mail.yahoo.de
|
