https://bz.apache.org/bugzilla/show_bug.cgi?id=69580

--- Comment #12 from Stefan Eissing <[email protected]> ---
Hi Jeff,

the patch from r1924267 remedies the situation that *if* a response is "headers
only" a missing EOS signal that is expected at the end of a response body, no
longer causes an error.

Usually, a response to a request signals 
  EOS (body ends) + EOR (request ends) + Flush. 
The trace you provided from mod_cache shows only 
  EOR + Flush.

which, before r1924267 led to a stream reset. Now, the 304 is passed and the
stream is closed. 

I did not see anything wrong in mod_cache regarding this. So, I assume this may
happen in other situations as well. In HTTP/1.1 it goes unnoticed. So HTTP/2 is
more "compatible" now.

The difference between h1 and h2 now is: if a 304 response would, as a bug,
really try to send body data, h2 would silently discard this, while in h1 it
would appear on the connection for a *subsequent* request.

So, there would be a scenario for request forgery here in h1, but not with h2
(before and after r1924267). I think r1924267 is good.

Thanks for checking r1924267.

-- 
You are receiving this mail because:
You are the assignee for the bug.
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]

Reply via email to