Re: sockets APIs extensions for Host Identity Protocol

[Keith Moore <[e-mail]>]

>> First is that there will always be some applications (such as
>> diagnostic tools) that need to know more about the network, so it is
>> necessary to have an API that allows those applications to function.
>>     
>
> We all (I believe) agree that we need a low-level API, mostly, as you
> said, for debugging / monitoring / teaching purposes and for some
> optimizations in very specific applications.
>
> But this is not the point to discuss since we *already* have this API,
> the sockets API, and it works.
>   
the sockets API is not sufficient even as a low-level API.  we're going
to need extensions to both network protocols and the sockets API to
function effectively in a world where hosts have multiple addresses and
it matters which address you use.
> What we discuss is a *new* API, a high-level one, not meant to replace
> the low-level one for *everything* but for most applications.
>   
we do need a high-level API, but such an API still needs to be able to
accept application preferences and requirements for that particular socket.

I generally find that assumptions about "most applications" turn out to
be wrong - if not in the present, than in the in the near future.
>> Fourth, for a variety of reasons, DNS names are not and have never
>> been adequate as general purpose endpoint identifiers,
>>     
>
> This is a sensible explanation but, if we agree with it, it only means
> that we rule out DNS for the identification of the new high-level
> "connection objects". This does not mean that the whole idea of a
> high-level API is wrong.
>   
the idea is fine, but the devil is in the implementation details.  it's
easy to design a high-level API that works for a limited number of
cases, very difficult to make one that works well in practice for
everyone in the diversity of the Internet.
> I use email. I may switch to SIP or Jabber. I did not find a RFC
> describing Skype.
>   
I'm quite willing to learn lessons from their example even if they don't
document their protocol in an RFC.

Keith