--- AnyEvent-HTTP/Changes 2010/06/16 19:17:30 1.37 +++ AnyEvent-HTTP/Changes 2011/01/02 04:50:40 1.54 @@ -2,11 +2,49 @@ TODO: provide lwp_request function that takes an lwp http requets and returns a http response. TODO: http://efl20.com/wp-register.php refresh-header, instead of location -TODO: GET => POST like mozilla does it, not like the rfc does it? -TODO: maybe support broken Expires header? +TODO: download example + + - hopefully fully upgraded to HTTP/1.1. + - use more differentiated status 59x codes. + - ignore leading zeroes in http version. + - much improved cookie implementation: + - properly implement cookie expiry (for new cookies). + - new function to expire cookies and sessions: cookie_jar_expire. + - add special exception to parse broken expires= keys in + set-cookie headers. + - do not quote cookie values when not strictly necessary, to + improve compatibility with broken servers. + - accept and send lots of invalid cookie values exactly as + they were received - this should not impact valid values. + - lowercase cookie parameter names for improved compatibility. + - support the max-age cookie parameter, used if expires is missing. + - support cookie dates (and a few others) in parse_date. + - suppoer value-less parameters (e.g. secure, httponly). + +1.5 Fri Dec 31 04:47:08 CET 2010 + - bugfix: after headers were received, if any error occured the wrong + (server-sent) Status and Reason fields would be passed to the callback. + - when an error occurs during transfer, preserve status/reason. + - add socks4a connect example. + - new "tcp_connect" parameter. + - new format_date and parse_date functions. + - diagnose unexpected eof as such when the length is known. + - add 205 to the responses without body. + +1.46 Mon Sep 6 08:29:34 CEST 2010 + - some (broken) servers differentiate between empty search parts + and nonexistant search parts, work around this (problem + analyzed by Sergey Zasenko). + - possibly increase robustness by always setting an on_error + callback on the AnyEvent::Handle object (especially in case + of user errors, such as nehative timeouts). + - we now always follow 301/302/303 redirects and mutate POST to GET. + - we now always follow 307 redirects, even for POST. + - header-less responses are not parsed correctly (at a negative + speed penatly :). 1.45 Wed Jun 16 21:15:26 CEST 2010 - - fix a bug where the handle woudl go away directly after a successful + - fix a bug where the handle would go away directly after a successful connect (analyzed and patch by Maxim Dounin). - due to popular demand, introduce the Redirect pseudo response header. - document URL pseudo-header better.