On 05/14/2014 07:48 PM, Pawel Veselov wrote: > On Wed, May 14, 2014 at 4:56 AM, Christian Grothoff > <groth...@gnunet.org>wrote: > >> On 05/14/2014 12:43 PM, Pawel Veselov wrote: >>> /me shamefully found >>> http://lists.gnu.org/archive/html/libmicrohttpd/2012-04/msg00002.html >>> >>> Is it then generally safe to assume that with "POST" method, I can always >>> reply with MHD_YES, expecting that access handler will be called again, >> or >>> do I need to check for crazy agents that may post with a 0 >> content-length? >> >> It is always safe to return MHD_YES. MHD_NO must only be returned if the >> connection must be closed due to a hard error, no HTTP error code is >> generated. >> > > I'm sorry, I meant "safe to just return MHD_YES", implying that whatever > processing I need to do for the response can be delayed, > because I'm guaranteed to being called back at least one more time.
Yes, that's correct. > If method is "POST", and context is 0, create context > Else, if *upload_bytes is 0, prepare for end of request, create status > Else, do whatever you need to do for incoming data > Return MHD_YES Right, that's the basic logic.
