Re: Protocols/APIs and redirects

On Wed, 07 Dec 2011 12:57:48 +0100, Julian Reschke <julian.reschke@gmx.de>  
wrote:
> My understanding is that it depends, and needs to decided on a  
> per-header basis. We can try classifications that make it easier to  
> decide, and we may even be able to recommend a default, but this will  
> break when a new header needs the non-default behavior.

I don't think that will work for us. It needs to be the same irrespective  
of the header. Or some kind of fixed unchanging list for which things is  
different.


> The base issue is splitting the responsibilities between two layers, and  
> have the lower layer (XHR) trying to decide things that the upper layer  
> (the script) should know.
>
> I'm not sure what this has to do with "HTML Fetch", as the problem is  
> specific to XHR. I recommend to fix the base issue first, which is that  
> clients can't ask XHR not to follow redirects.

XMLHttpRequest uses HTML fetch. Server-sent events does too, and a number  
of other things affected do so as well. You indeed keep bringing up that  
we should add a feature to XMLHttpRequest and that it will resolve the  
problem. But adding a feature does not resolve the problem. We will still  
have the problem for when such a feature is not used. So lets address  
that. Adding a feature for redirects is completely orthogonal as I  
explained several times now. I am getting somewhat tired of having to  
repeat myself.


-- 
Anne van Kesteren
http://annevankesteren.nl/