Re: [Ecrit] draft-ietf-ecrit-lost-sync-15.txt

[Hannes Tschofenig <hannes.tschofenig at gmx.net>]

Hi Julian, 

On Jan 15, 2012, at 6:09 PM, Julian Reschke wrote:

>> There are two interesting issues:
>> 
>> 1) HTTP POST is used by LoST Sync
>> 
>> When Henning and I worked on the LoST Sync specification we just followed the design of many of our other HTTP protocol designs (like LoST, HELD, etc.).
> 
> Yep.
> 
> The protocols mentioned above essentially use HTTP as a tunnel. I would NOT recommend them as an example of protocol design unless you are sure that you can control senders, receivers, and intermediates. Also keep in mind that one of the advantages of using HTTP is that you can reuse existing components (like libraries and server frameworks), in which case using HTTP the wrong way can lead to later pain.

But didn't we add directives that control the behavior of the HTTP protocol to, for example, control caching? 
Isn't this enough or the wrong approach? 

Ciao
Hannes