Richard wrote:
>> except the httpconnection implementations on most of the linux versions suck :( >failure to timeout properly is one common bug. >> Actually, this bug applies to Java in general. For some reason the generic design of the URLConnection class does not permit implementors to set socket timeouts. Annoying but true. But with some good will and subclassing this can be worked around. (Or so I've heard, I've not tried it myself). Ed wrote: >> The obvious next question is, how scalable is it in terms of high-volume >applications? IOW, how many page requests/sec would it support? I'm sure there is >an upper limit where C/C++ coding would better fit the need. Or is there? >> No, or at least not in the sense I meant "scalability". Remember that Java is designed with networking in mind. The "JavaSpace" for example can be described as a networkable combination of a message queue and a object database. It supports atomic operations and transactions and can be used transparent over a network. So an application can be designed in "services" where each service is a standalone server executing parts of the application task and communicating with the other services through JavaSpaces. Since the JavaSpace operations are atomic and supports transactions it's easy to scale since the only thing you'd have to do (provided you did the original design right) is to plug in another server. Yours - Lars J. Nilsson [EMAIL PROTECTED] -- This message was sent by the Internet robots and spiders discussion list ([EMAIL PROTECTED]). For list server commands, send "help" in the body of a message to "[EMAIL PROTECTED]".
