Re: Making extensibility cheap

On 4/06/2014 5:32 p.m., Mike Bishop wrote:
> The second is what's in my draft -- I just copied your proposed version of DATA, added some extra flags, and called it a COMPRESSED_DATA frame.  I think it's hard to sell it as an end-to-end frame for a couple reasons, but the biggest is that you don't have the opportunity for negotiation with the far side.  That's why I keep coming back to an end-to-end frame that can be down-converted to a DATA frame if it hits a hop that doesn't support the extension.
> 

I agree. The end-to-end is far better serviced by content-encoding
already. A hop-by-hop replacement for transfer-encoding in an extension
RFC would be great and does not need to hold up the core spec document.

Amos

Received on Wednesday, 4 June 2014 11:05:35 UTC