Re: last example on 8.1.1 is faulty?

On 5 October 2013 08:31, Salvatore Loreto <salvatore.loreto@ericsson.com> wrote:
> The TE header is a request header. It should probably be Transfer-Encoding:
> chunked.

TE: trailers is used solely to signal the presence of trailers.

Transfer-Encoding is absent, yes.  That's an omission.

> Also content-length and Transfer-Enconding are not use at the ame time.

I'm not aware of a mutual exclusion clause, but I'm not looking very
hard right now.  Content-Length becomes merely metadata in both cases.

> It is a little bit unclear how indication of content length (closing the
> connection, content-length
> or transfer encoding: chunked as in HTTP/1.0 and 1.1) relates to the
> END_STREAM and END_HEADERS flags.

The last DATA frame on a stream terminates the payload.

> For dynamically generated content, content-length is not always known, so
> can content-length be used
> in this particular example?

We could skip it, but what value does that provide?

Received on Monday, 7 October 2013 22:01:26 UTC