Depends how you use it. For buffering realtime data if latency is of chief concern then by all means, discard the old stuff. If the desired behavior is to block when the queue is full, then you need to check for an overlap.

If you're overwriting old data you still need to check for overlap and advance the read index if you're overrunning old data.

Thanks, this is what I was missing, and seems right in general. I think the old ring buffers I'm remembering just decided that reading out-of-order was a fitting punishment for not reading often enough. After all there is a bit of a race for the read index here...

I can't read the site because it's blocked for me, but isn't a ring-buffer a different class of data-structure than a queue? AIUI, queues are for non-lossy FIFO behavior.

No, ring buffers are just an implementation of a queue. You can implement lossy/lossless ring buffers or linked-list queues.

My $10/mo shared host is already spinning with ycomb+linkedin traffic. I shall upgrade...