Re: END_SEGMENT and headers, #2

On 2014–04–19, at 4:23 AM, Martin Thomson <martin.thomson@gmail.com> wrote:

> On 18 April 2014 13:12, David Krauss <potswa@gmail.com> wrote:
>> Not to beat a dead horse, but I think that bit should go. Data frames alone provide exactly enough END_SEGMENT flags.
> 
> Roberto described why he wanted it: to be able to finish a segment
> with a HEADERS frame.  I don't see how you can address that need
> without the bit.


With an empty DATA frame, which is exactly how I’m doing my interface. But it really sounds like a corner case. Metadata is much more likely to occur at the beginning of a new message than later, or especially the end.

I can see translating HTTP-like transactions to segments, and you would END_SEGMENT after each response. But sending an empty data frame for an empty response body is only 8 bytes. In the old currency that only buys you the string "HTTP/1.1” :v) .