Soon back (i think)
Project I'm currently working on, ends in ~1month. (deadline)
Time I currently have for coding io (15mins-1hour / day), I'm spending on designing new core.
Faster core I'm describing, will use somewhat more memory per connection (as each connection will have dynamic buffers) But it should reduce 'buffer copying' to near zero, whenever cpu is not the bottle neck for network related io operations. (implementing it, once ready - should be quite easy, as all transfers in io nowdays rely on TransmitBuffers() api command :-))
What comes to io running as NT-service, it's all planned with notepad ;-p If all goes accoring to plan, io will have *nix like 'init' levels before summer. (go figure)
|