On May 1, 2014:11:02 AM, at 11:02 AM, Joe Touch <touch(_at_)isi(_dot_)edu>
wrote:
On 5/1/2014 5:12 AM, Thomas Nadeau wrote:
APIs are not that useful unless there is code behind them.
Ultimately, yes. But the code represents an instance of the API.
That depends on your perspective. These days the code IS the API, in
particular open source code. Standards bodies do not need to define the APIs;
implementation communities do that already. The IETF should probably stick to
on-the-wire protocols.
The "application" layer actions in RFC793 - SEND, RECEIVE, CONNECT, LISTEN -
are *not* the same as the Unix socket API; Unix sockets are one
implementation of that interface.
Are you proposing code that goes with those APIs?
Yes, just as I would propose code that implements a protocol. But I don't
think either one is useful in the IETF except as an example - definitely
never as a specification.
Also the language that you do this in is important depending on the
area of applicability.
Just as much (or little) as the physical layer (802.11, ethernet, carrier
pigeon) is important to IP.
For instance, if these APIs are related to modern
applications, C is all but useless because its not used to build most of
those; you need Java/python/etc... for these cases.
If you don't have C, you often don't have scripted languages that are (often)
compiled from C source code.
However, that depends on your compiler and the environment in which you
develop your languages. That's important for language developers, but not the
IETF (any more than we spec how to build pigeon coops).
This is why I personally think this sort of effort would not be very
useful; modern applications are simply not built this way any longer.
--Tom
signature.asc
Description: Message signed with OpenPGP using GPGMail