Representation = entity in HTTPbis

Is anyone else on or near the TAG monitoring HTTPbis? I accidentally  
found a proposal [1,2] from last July, which would redefine  
'representation' to include things like the payload of a 300, 404, or  
501 response, or a PUT or POST request. That would be at variance both  
with REST (Fielding ICSE 2000) and AWWW, and with the literature and  
specs that have followed in their wake.

I've posted a comment [2] to ietf-http-wg@w3.org to this effect.

There should be some kind of separation of concerns between HTTP and  
webarch, or at least coordination. In this case maybe the solution  
would be to remove the word 'representation' from HTTP.

Of course this is not a disaster, since the work is still in progress,  
and dissonances like this can be remedied during pubic review of  
HTTPbis. I just want to alert people that there are actions afoot that  
might make the public even more confused than it already is.

I expected a correction from Roy Fielding, but it probably went right  
past him. I'll keep an eye on HTTPbis from time to time, but anyone  
else who cares about spec harmony is advised to do the same.

I've already alerted HTTPbis to disagreement between 2616 and 3986  
over the meaning of 'resource' and am now on the hook to propose a fix.

Jonathan

[1] http://trac.tools.ietf.org/wg/httpbis/trac/ticket/109
[2] http://www.w3.org/mid/BD20ED9D-D864-4DA8-B1E8-26276B237FBC@mnot.net
[3] http://lists.w3.org/Archives/Public/ietf-http-wg/2009JanMar/0182.html

Received on Friday, 6 February 2009 13:07:33 UTC