Re: [SRI] What should we Hash Redux

I'm not sure why you refer to Fetch as SW. There's only a tiny part of
Fetch that defers SW, otherwise it mostly defines the platform's
URL/networking stack.

On Thu, Jul 3, 2014 at 6:37 PM, Devdatta Akhawe <dev.akhawe@gmail.com> wrote:
>> Per HTTP the payload body is a message body with any content codings removed.
>
> See mnot's note:
> http://lists.w3.org/Archives/Public/public-webappsec/2014Mar/0026.html
>
> Payload removes gzip transfer-encodings but not content encoding.
> Based on the thread, it seemed like there was no simple "spec

The spec seems to suggest there is:

https://tools.ietf.org/html/rfc7230#section-3.3


>> the platform today, including XMLHttpRequest, expose the payload body.
>
> Per above, are you referring here to "with gzip content encoding
> removed" or without?

With.


>> fetch() exposes the message body (as a stream, though the only methods
>> available on that stream undo the content codings and give you the
>> payload body as a result).
>
> So, fetch, XHR both use body with content codings removed? What
> happens when it is a tar.gz file with Content-Encoding: gzip? Does
> fetch and XHR remove the codings?

No, XMLHttpRequest, <img>, and such do, fetch() does not.


>> <a download> is likely not fully defined.
>> From what I understood from bz it will depend on what is being
>> downloaded and what the file extension situation looks like...
>
> yeah.  SW can interact with a download too, right? What will that look like?

It seems HTML does not define this in detail at the moment. That would
need to be fixed.


-- 
http://annevankesteren.nl/

Received on Thursday, 3 July 2014 16:50:32 UTC