[Live-devel] Performance/Security concern

[Ross Finlayson]

>However, in studying the code to learn about the architecture, I
>encountered something odd: The RTSP connection is blocking. In
>RTSPServer::RTSPClientSession::incomingRequestHandler1(), the code
>essentially waits until a full RTSP request header has arrived. As a
>result, if I connect to the server and begin sending a header, but do
>not finish the header, the server is wedged until I either complete the
>header or disconnect.

Yes - good call.  You're correct - to fit the code's event-driven 
execution model, all of the server's request-handling reads (not just 
the first one), should be done asynchronously.

I will be shortly be releasing a new version of the code to fix this 
problem.  Stay tuned...

(It turns out that "RTSPClient" has the same problem, but that's less 
serious - for most applications, it doesn't matter so much if a RTSP 
client request blocks.)
-- 

Ross Finlayson
Live Networks, Inc.
http://www.live555.com/