The purpose of HttpNIO (the way I see it) is to provide a non-blocking version of core HTTP components based exactly on the select/callback architecture
Oh! Well, I suppose I'll still work on using HttpAsync to make an HttpNB. I very much like the pattern it provides for dispatching and handling requests.
There is one important word missing :-) weeks? months?
Heh. I meant to say months. A better estimate might be three months. It depends on how many unexpected bugs crop up with the download conversion & how much extra work is required in separating out the i/o package. The good knews is that once that part is finished, the code will have been well tested on millions of machines in the world, so there should be very little bugs effecting HttpNB.
HttpAsync is in a very early stage. It reflects my attempt to understand the problem from a purely blocking IO perspective. If you feel that some operations should be factored out, feel free to submit patches. Or, when I'm back to coding, suggestions for patches.
Yup. I had planned to submit some patches and suggestions once I got around to the actual task of working with HttpClient. It felt a bit premature to suggest changes at this early stage.
Once I've got some design documentation on the HttpAsync interface and initial blocking implementation, I suggest you add a section on the design for a non-blocking implementation
That sounds good. I'll try to make a pass at that after looking at your documentation. Thanks, Sam
