aboutsummaryrefslogtreecommitdiffstats
path: root/src
AgeCommit message (Collapse)Author
2011-05-14Add the required Date header in the HTTP replies.Loïc Hoguin
The formatted date is generated and kept up to date regularly by a gen_server process storing it in the cowboy_clock ets table. Then it is retrieved by other processes simply by reading the table.
2011-05-09Add a max_connections transport option.Loïc Hoguin
Limits the number of parallel requests processed at once. Waiting requests are kept in the accept queue. The limit is not explicitly observed, but it should be around the given value at any time. Defaults to 1024. Thanks to ostinelli's benchmark to point out the issue hopefully solved by this and the previous commit.
2011-05-09Increase the default backlog from 128 to 1024.Loïc Hoguin
Improves reliability under heavy load.
2011-05-08Add chunked reply support.Loïc Hoguin
Send the status line and headers using cowboy_http_req:chunked_reply/3, and individual chunks with cowboy_http_req:chunk/2.
2011-05-08Separate building the response head from replying.Loïc Hoguin
2011-05-06Allow atoms for response header namesAnthony Ramine
2011-05-05Switch the HTTP protocol to use binary packets instead of lists.Loïc Hoguin
The server now does a single recv (or more, but only if needed) which is then sent to erlang:decode_packet/3 multiple times. Since most requests are smaller than the default MTU on many platforms, we benefit from this greatly. In the case of requests with a body, the server usually read at least part of the body on the first recv. This is bufferized properly and used when later retrieving the body. In the case of pipelined requests, we can end up reading many requests in a single recv, which are then handled properly using only the buffer containing the received data.
2011-05-04Add cowboy_http_req:port/1.Loïc Hoguin
Returns the port given in the Host header if present, otherwise the default port of 443 for HTTPS and 80 for HTTP is returned.
2011-05-04Use http_headers() type in cowboy_http_req:headers/1.Loïc Hoguin
2011-05-03Add cowboy_sup to the list of registered processesAnthony Ramine
Cowboy top supervisor has a name and it should be clearly stated in its .app.src file.
2011-04-30Add backlog transport option.Loïc Hoguin
This fixes issues with the http_load benchmark tool. The default backlog option from OTP only queues up to 5 connections, which is way too low for a fast-responding server. Issue initially found thanks to DeadZen bugging me to test cowboy with http_load. Fix found thanks to ostinelli's misultin already having the backlog option which was the one thing it did differently than cowboy.
2011-04-30Remove the next_request function always matching its first clause.Loïc Hoguin
2011-04-29Fix type specs for ssl:socket.Loïc Hoguin
Should fix the warnings given by the Dialyzer-current build over at http://dialyzer.softlab.ntua.gr/current/cowboy/ Thanks Kostis for the help.
2011-04-18Remove the include/types.hrl file.Loïc Hoguin
2011-04-18Replace port_number() by ip_port() from kernel/include/inet.hrl.Loïc Hoguin
2011-04-18Replace posix() by atom(), for now.Loïc Hoguin
The type isn't exported by OTP so we don't win much. Also, inet.erl and file.erl define posix() différently, so OTP needs to stop being so confuse in the first place before we can attempt to use it properly.
2011-04-18Move the type application_start_type() into cowboy_app.Loïc Hoguin
2011-04-18Move the dispatcher related types into cowboy_dispatcher.Loïc Hoguin
2011-04-18Rename match() into match_rule().Loïc Hoguin
2011-04-18Rename dispatch() into dispatch_rules().Loïc Hoguin
2011-04-18Move the opaque type sslsocket() out of include/types.hrl.Loïc Hoguin
2011-04-18Use the inet:socket() type instead of the user-defined one.Loïc Hoguin
2011-04-17Use the inet:ip_address() type instead of the user-defined one.Loïc Hoguin
2011-04-16Add an accept timeout for code reloading.Loïc Hoguin
Thought it was already there, but I guess not. Anyway it's here now so everything is back to normal. ;)
2011-04-14SSL transport uses sslsocket(), not plain socket().Loïc Hoguin
2011-04-14Initial draft-hixie-thewebsocketprotocol-76 support.Loïc Hoguin
2011-04-14Fix specs for cowboy_http_req:reply/4.Loïc Hoguin
Body is iodata(), not iolist().
2011-04-14Fix a bug where dupe headers were sent in cowboy_http_req:reply/4.Loïc Hoguin
Now the server defines default headers that can be overwritten by the handler simply by passing them to the reply/4 function. Default headers include, for now, Connection and Content-Length headers. Note that it isn't enough to change the Connection header to close a keep-alive connection server-side.
2011-04-14Add a test case for cowboy_dispatcher:match/3.Loïc Hoguin
Added it while debugging an issue, so why not keep it?
2011-04-12Add a Transport:messages/0 function to obtain the atoms sent in active mode.Loïc Hoguin
That way we can support active mode without having to hardcode the atoms for each transports available.
2011-04-12Fix a pattern matching bug in cowboy_http_protocol:handler_init/2.Loïc Hoguin
2011-04-09Limit the number of empty lines to allow before the request-line.Loïc Hoguin
Defaults to 5. Prevents someone from indefinitely sending empty lines.
2011-04-08Discard the port from the host before tokenizing it.Loïc Hoguin
2011-04-07Don't crash on errors while receiving headers; throw an error 500 instead.Loïc Hoguin
The server shouldn't crash the request process when we have an error while receiving headers. A case where this could happen is if the header line is too long. See also bfrog's report on ticket #3 on github.
2011-04-05Don't automatically retrieve the peer information for the 'OPTIONS' method.Loïc Hoguin
2011-04-05Handle properly the default Connection value for HTTP/1.0.Loïc Hoguin
Fixes a bug reported by evaxsoftware on IRC.
2011-04-03Add initial SSL transport support.Loïc Hoguin
Code is working but hasn't been benchmarked yet.
2011-04-03Don't ignore the return values in cowboy:stop_listener/1.Loïc Hoguin
2011-04-03Ensure the listening socket gets closed with the listener supervisor.Loïc Hoguin
Moved the Transport:listen call to cowboy_acceptors_sup. This make it depend on a child of cowboy_listener_sup instead of cowboy_sup, which isn't getting shut down when calling cowboy:stop_listener/1 and thus didn't close the listening socket.
2011-04-03Make Cowboy an OTP application again, properly this time.Loïc Hoguin
As requested by many people on IRC Cowboy is now a proper OTP application to support soft code upgrades. It should also be easier to start and stop listeners now using cowboy:start_listener/6 and cowboy:stop_listener/1.
2011-04-02Anonymize and improve the cowboy supervision tree.Loïc Hoguin
* Cowboy isn't an OTP application anymore; just a supervisor. * All processes started by Cowboy are now anonymous. * All processes related to a listener are now part of its supervision tree.
2011-03-29Make the default value the last parameter of HTTP req accessorsAnthony Ramine
This makes cowboy_http_req's qs_val/3, binding/3 and header/3 behave like proplists:get_value/3.
2011-03-29Make HTTP req accessors call themselves with their default default valueAnthony Ramine
Use lists:keyfind/3 in qs_val/3, binding/3 and header/3 and make qs_val/2, binding/2 and header/2 call those.
2011-03-29Fix cowboy_http_req:body with regard to the new accessors default valueAnthony Ramine
header/2 now returns 'undefined' when the header isn't in the request.
2011-03-27Return undefined instead of failing or returning "" when a value doesn't exist.Loïc Hoguin
2011-03-27Use a more efficient variant of string:to_lower to improve performance.Loïc Hoguin
After much testing and experimentation of all kinds I find lists to be both faster and using less memory than binaries for request-line and headers handling. This is more than likely due to the fact that headers are very short and thus do not benefit from the advantages of refc binaries, meaning they're copied, just like lists. The memory usage discrepancy is still a mystery for the most part, although the hoops needed to perform operations on the binaries are probably responsible for the extra memory use. I'm thus giving up on trying to use binaries for request-line and headers. Instead, this commit improves performances even more to the lists code, making lists 5% faster than binaries. Lists are easier to work with too, so I guess it's all a big win for everyone. Of course the request body is still read as a binary, we're using the binary type where it performs best.
2011-03-27Match early in connection_to_atom for a small performance improvement.Loïc Hoguin
2011-03-27Fix query string related type specs in cowboy_http_req.Loïc Hoguin
2011-03-24Add the cowboy_http_handler behaviour.Loïc Hoguin
2011-03-22Allow Handler:init/3 to request a protocol upgrade.Loïc Hoguin