On Tue, Oct 21, 2008 at 3:32 PM, newToMina <[EMAIL PROTECTED]> wrote: > > Ok. If I understood correctly, as long as the clients write one xml document > at a time to the socket, then most of problems should go away, correct? If > that is the case, the only problem (or are there more?) I could see is in > the event of one message spanning over multiple reads. If I extend the > TextLineDecoder and read until it find the end root tag shouldn't it take > care of that problem too?
One xml document written in a socket could mean several reads on the other side, and could (potentially at least) take minutes, hours, days to arrive, so you can't rely on that. Can't you add a header and footer to the message you send? Something like --header--<xml ...>...</xml>--end-- where --header-- and --end-- may be a String or a single byte indicating the end of the document
