On Fri, 13 Jan 2006, Duane Wessels wrote:
I was bothered that ConnStateData has this:struct { size_t size_left; /* How much body left to process */ HttpRequest *request; /* Parameters passed to clientReadBody */ char *buf; size_t bufsize; CBCB *callback; void *cbdata; } body; which is pretty similar to the way ClientBody looked: private: ConnStateData::Pointer conn; char *buf; CBCB *callback; HttpRequest *request; }; So I basically merged the two. ConnStateData will allocate one ClientBody if necessary during request parsing, instead of on a per-call basis.
Looks fine to me. Finishes the transition from being embedded in ConnStateData into more complete isolation of request entity processing in it's own class ClientBody.
Regarding the duplicated entries it looks to me as this transition was started but not completely finished yet. With your changes the transition is completed and very clean. Even easy to follow. Good work, assuming it works :-)
Regards Henrik
