On Oct 18, 2010, at 10:03 PM, Levente Uzonyi wrote:

On Mon, 18 Oct 2010, Stéphane Ducasse wrote:

...

...

This benchmark is the real concurrent stress test. 100 processes are adding 10,000 elements to the queue while another 100 are reading from it. It clearly shows that Igor's queues are an order of magnitude faster. Also 200 concurrent processes cause much less slowdown compared to the sequential tests for them.

So, even though SharedQueue is now faster than SharedQueue2, both will have to go IMHO. :)

Hi Levente

Naively why?

Did you see the benchmarks?

...

and replace by what?

By Igor's new collections. Though the name should be kept for compatibility reasons.

Ok I was confused... too many emails and threads. So yes! Stef