Re: Quick feedback on draft-nottingham-web-proxy-desc-00

On Thu, 02 Oct 2014 18:20:02 +0200, Smith, Kevin, (R&D) Vodafone Group  
<Kevin.Smith@vodafone.com> wrote:

>>> Rightfully. This is what RFC 7230 already requires.
>
> Why rightfully, though? The example given has merit, and could make the  
> difference between a functional user experience, and not.

There are two relevant failure modes here. Something breaks because a  
proxy did something, or something breaks because the proxy didn't do  
something. The spec currently favors the second one, which arguably makes  
debugging simpler and failures easier to reproduce across different  
networks. E.g. instead of correcting PNG images with the old palette  
buffer overrun bug, you should block the delivery.

It would however have been good if the client was also part of this  
decision. E.g. sending no-transform or do-transform in the request. Also,  
I have yet to see a legitimate use for no-transform in a real world  
application.

/Martin Nilsson

-- 
Using Opera's mail client: http://www.opera.com/mail/

Received on Friday, 3 October 2014 00:20:42 UTC