Oh, I see what you mean. You're right. I don't know what I was thinking (not enough coffee yet or something). The BoundedBuffer wrapper would actually enforce a maximum size and wait until something was removed from it (there you'd have to also override the remove and clear methods and the iterator implementation I believe). So, the BoundedBuffer class would need a few settings:
boolean waitWhenFull; long maximumWait; int maximumSize; If waitWhenFull is false, then it just throws a BufferOverflowException (sounds so villainous). The maximumWait would default to -1 (wait forever). The TimoutBuffer wouldn't need to have a "waitOnEmpty" setting, since you'd just use the default implementation which would throw a BufferUnderflowException. We would need a maximumWait property on TimeoutBuffer, though (defaulting to -1 also). Does this sound right? -----Original Message----- From: Stephen Colebourne [mailto:[EMAIL PROTECTED] Sent: Thursday, November 10, 2005 7:40 AM To: Jakarta Commons Developers List Subject: Re: [collections] New Buffer Wrapper Classes... These sound like good ideas. Not sure why you need two TimeoutWrappers though, as you are only trapping the remove() aren't you? Stephen --- James Carman <[EMAIL PROTECTED]> wrote: > Based on my response to Sandy McArthur about using > Buffers to implements > Commons Pool, I started digging around the API. I > came up with a couple of > ideas: > > 1. A BoundedBuffer wrapper/decorator class, which > basically enforces a > maximum size of the enclosed buffer. > > 2. A TimeoutBuffer wrapper/decorator class which > waits until a specified > timeout value expires when asking to remove() > objects from an empty Buffer. > > With these two classes, we can make any underlying > implementation (stack, > queue, priority queue, etc.) bounded or cause them > to wait rather than throw > a BufferUnderflowException immediately when removing > objects from them. We > may have to come up with two timeout wrappers, one > for read and one for > write (optionally they could be in the same class > with potentially different > values for each). What do you guys think? --------------------------------------------------------------------- To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED] --------------------------------------------------------------------- To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED]
