Re: API, was: Question re HIP dependency & some architectural considerations

On 29-jul-04, at 13:32, Spencer Dawkins wrote:

> Agree on T/TCP, confused on HTTP persistent connections. I'm just
> caught on "badly implemented and extremely rarely used".

I mainly meant that HTTP persistence is also rarely used, I'm not to 
sure about the details. As for T/TCP, don't know about the security 
problems (and don't care for the most part, it's not like I'm going to 
use T/TCP for BGP sessions or anything) but the Linux or BSD 
implementation (I forget which) basically treats T/TCP as a single 
datagram in both directions on steroids rather than regular TCP that 
establishes very quickly, thereby making it impossible to use T/TCP in 
practice for many applications.

> Got a pointer to other current problems?

In this area or in general?

I just wish we could ditch IP altogether and start from scratch...