Feersum - A PSGI engine for Perl based on EV/libev


    use Feersum;
    my $ngn = Feersum->endjinn; # singleton
    # register a PSGI handler
    $ngn->psgi_request_handler(sub {
        my $env = shift;
        return [200,
            ["You win one cryptosphere!\n"]];
    # register a Feersum handler:
    $ngn->request_handler(sub {
        my $req = shift;
        my $t; $t = EV::timer 2, 0, sub {
                ['Content-Type' => 'text/plain'],
                \"You win one cryptosphere!\n"
            undef $t;


Feersum is an HTTP server built on EV. It fully supports the PSGI 1.03 spec including the psgi.streaming interface and is compatible with Plack. PSGI 1.1, which has yet to be published formally, is also supported. Feersum also has its own "native" interface which is similar in a lot of ways to PSGI, but is not compatible with PSGI or PSGI middleware.

Feersum uses a single-threaded, event-based programming architecture to scale and can handle many concurrent connections efficiently in both CPU and RAM. It skips doing a lot of sanity checking with the assumption that a "front-end" HTTP/HTTPS server is placed between it and the Internet.

How It Works

All of the request-parsing and I/O marshalling is done using C or XS code. HTTP parsing is done by picohttpparser, which is the core of HTTP::Parser::XS. The network I/O is done via the libev library. This is made possible by EV::MakeMaker, which allows extension writers to link against the same libev that EV is using. This means that one can write an evented app using EV or AnyEvent from Perl that completely co-operates with the server's event loop.

Since the Perl "app" (handler) is executed in the same thread as the event loop, one need to be careful to not block this thread. Standard techniques include using AnyEvent or EV idle and timer watchers, using Coro to multitask, and using sub-processes to do heavy lifting (e.g. AnyEvent::Worker and AnyEvent::DBI).

Feersum also attempts to do as little copying of data as possible. Feersum uses the low-level writev system call to avoid having to copy data into a buffer. For response data, references to scalars are kept in order to avoid copying the string values (once the data is written to the socket, the reference is dropped and the data is garbage collected).

A trivial hello-world handler can process in excess of 5000 requests per second on a 4-core Intel(R) Xeon(R) E5335 @ 2.00GHz using TCPv4 on the loopback interface, OS Ubuntu 6.06LTS, Perl 5.8.7. Your mileage will likely vary.

For even faster results, Feersum can support very simple pre-forking (See feersum, Feersum::Runner or Plack::Handler::Feersum for details).


There are two handler interfaces for Feersum: The PSGI handler interface and the "Feersum-native" handler interface. The PSGI handler interface is fully PSGI 1.03 compatible and supports psgi.streaming. The psgix.input.buffered and features of PSGI 1.1 are also supported. The Feersum-native handler interface is "inspired by" PSGI, but does some things differently for speed.

Feersum will use "Transfer-Encoding: chunked" for HTTP/1.1 clients and "Connection: close" streaming as a fallback. Technically "Connection: close" streaming isn't part of the HTTP/1.0 or 1.1 spec, but many browsers and agents support it anyway.

Currently POST/PUT does not stream input, but read() can be called on psgi.input to get the body (which has been buffered up before the request callback is called and therefore will never block). Likely read() will change to raise EAGAIN responses and allow for a callback to be registered on the arrival of more data. (The psgix.input.buffered env var is set to reflect this).

PSGI interface

Feersum fully supports the PSGI 1.03 spec including psgi.streaming.

See also Plack::Handler::Feersum, which provides a way to use Feersum with plackup and Plack::Runner.

Call psgi_request_handler($app) to register $app as a PSGI handler.

    my $app = do $filename;

The env hash passed in will always have the following keys in addition to dynamic ones:

    psgi.version      => [1,0],
    psgi.nonblocking  => 1,
    psgi.multithread  => '', # i.e. false
    psgi.multiprocess => '',
    psgi.streaming    => 1,
    psgi.errors       => \*STDERR,
    SCRIPT_NAME       => "",

Feersum adds these extensions (see below for info)

    psgix.input.buffered   => 1,
    psgix.output.buffered  => 1,
    psgix.body.scalar_refs => 1,
    psgix.output.guard     => 1,               => \$magical_io_socket,

Note that SCRIPT_NAME is always blank (but defined). PATH_INFO will contain the path part of the requested URI.

For requests with a body (e.g. POST) psgi.input will contain a valid file-handle. Feersum currently passes undef for psgi.input when there is no body to avoid unnecessary work.

    my $r = delete $env->{'psgi.input'};
    $r->read($body, $env->{CONTENT_LENGTH});
    # optional: choose to stop receiving further input, discard buffers:

The psgi.streaming interface is fully supported, including the writer-object poll_cb callback feature defined in PSGI 1.03. Note that poll_cb is removed from the preliminary PSGI 1.1 spec. Feersum calls the poll_cb callback after all data has been flushed out and the socket is write-ready. The data is buffered until the callback returns at which point it will be immediately flushed to the socket.

    my $app = sub {
        my $env = shift;
        return sub {
            my $respond = shift;
            my $w = $respond->([
                200, ['Content-Type' => 'application/json']
            my $n = 0;
            $w->poll_cb(sub {
                # will also unset the poll_cb:
                $_[0]->close if ($n++ >= 100);

Note that $w->close() will be called when the last reference to the writer is dropped.

PSGI extensions


Scalar refs in the response body are supported, and is indicated as an via the psgix.body.scalar_refs env variable. Passing by reference is significantly faster than copying a value onto the return stack or into an array. It's also very useful when broadcasting a message to many connected clients. This is a Feersum-native feature exposed to PSGI apps; very few other PSGI handlers will support this.


Calls to $w->write() will never block. This behaviour is indicated by psgix.output.buffered in the PSGI env hash.


psgix.input.buffered is defined as part of PSGI 1.1. It means that calls to read on the input handle will never block because the complete input has been buffered in some way.

Feersum currently buffers the entire input in memory calling the callback.

Feersum's input behaviour MAY eventually change to not be psgix.input.buffered! Likely, a poll_cb() method similar to how the writer handle works could be registered to have input "pushed" to the app.


The streaming responder has a response_guard() method that can be used to attach a guard to the request. When the request completes (all data has been written to the socket and the socket has been closed) the guard will trigger. This is an alternate means to doing a "write completion" callback via poll_cb() that should be more efficient. An analogy is the "on_drain" handler in AnyEvent::Handle.

A "guard" in this context is some object that will do something interesting in its DESTROY/DEMOLISH method. For example, Guard.

The raw socket extension is provided in order to support Web::Hippie and websockets. is defined as part of PSGI 1.1. To obtain the IO::Socket corresponding to this connection, read this environment variable.

The underlying file descriptor will have O_NONBLOCK, TCP_NODELAY, SO_OOBINLINE enabled and SO_LINGER disabled.

PSGI apps MUST use a psgi.streaming response so that Feersum doesn't try to flush and close the connection. Additionally, the "respond" parameter to the streaming callback MUST NOT be called for the same reason.

    my $env = shift;
    return sub {
        my $fh = $env->{''};
        syswrite $fh, 

The Feersum-native interface

The Feersum-native interface is inspired by PSGI, but is inherently incompatible with it. Apps written against this API will not work as a PSGI app.

This interface may have removals and is not stable until Feersum reaches version 1.0, at which point the interface API will become stable and will only change for bug fixes or new additions. The "stable" and will retain backwards compatibility until at least the next major release.

The main entry point is a sub-ref passed to request_handler. This sub is passed a reference to an object that represents an HTTP connection. Currently the request_handler is called during the "check" and "idle" phases of the EV event loop. The handler is always called after request headers have been read. Currently, the handler will only be called after a full request entity has been received for POST/PUT/etc.

The simplest way to send a response is to use send_response:

    my $req = shift;
    $req->send_response(200, \@headers, ["body ", \"parts"]);

Or, if the app has everything packed into a single scalar already, just pass it in by reference.

    my $req = shift;
    $req->send_response(200, \@headers, \"whole body");

Both of the above will generate Content-Length header (replacing any that were pre-defined in @headers).

An environment hash is easy to obtain, but is a method call instead of a parameter to the callback. (In PSGI, there is no $req object; the env hash is the first parameter to the callback). The hash contains the same items as it would for a PSGI handler (see above for those).

    my $req = shift;
    my $env = $req->env();

To read input from a POST/PUT, use the psgi.input item of the env hash.

    if ($req->{REQUEST_METHOD} eq 'POST') {
        my $body = '';
        my $r = delete $env->{'psgi.input'};
        $r->read($body, $env->{CONTENT_LENGTH});
        # optional:

Starting a response in stream mode enables the write() method (which really acts more like a buffered 'print'). Calls to write() will never block.

    my $req = shift;
    my $w = $req->start_streaming(200, \@headers);
    $w->write(\"this is a reference to some shared chunk\n");
    $w->write("regular scalars are OK too\n");
    $w->close(); # close off the stream

The writer object supports poll_cb as also specified in PSGI 1.03. Feersum will call the callback only when all data has been flushed out at the socket level. Use close() or unset the handler ($w->poll_cb(undef)) to stop the callback from getting called.

    my $req = shift;
    my $w = $req->start_streaming(
        "200 OK", ['Content-Type' => 'application/json']);
    my $n = 0;
    $w->poll_cb(sub {
        # $_[0] is a copy of $w so a closure doesn't need to be made
        $_[0]->close if ($n++ >= 100);

Note that $w->close() will be called when the last reference to the writer is dropped.


These are methods on the global Feersum singleton.


Returns the Feersum singleton. Takes no parameters.


Use the file-descriptor attached to a listen-socket to accept connections.

TLS sockets are NOT supported nor are they detected. Feersum needs to use the socket at a low level and will ignore any encryption that has been established (data is always sent in the clear). The intented use of Feersum is over localhost-only sockets.

A reference to $sock is kept as Feersum->endjinn->{socket}.


Use the specified fileno to accept connections. May be used as an alternative to use_socket.


Stop listening to the socket specified by use_socket or accept_on_fd.

request_handler(sub { my $req = shift; ... })

Sets the global request handler. Any previous handler is replaced.

The handler callback is passed a Feersum::Connection object.

Subject to change: if the request has an entity body then the handler will be called only after receiving the body in its entirety. The headers *must* specify a Content-Length of the body otherwise the request will be rejected. The maximum size is hard coded to 2147483647 bytes (this may be considered a bug).

psgi_request_handler(sub { my $env = shift; ... })

Like request_handler, but assigns a PSGI handler instead.


Get or set the global read timeout.

Feersum will wait about this long to receive all headers of a request (within the tollerances provided by libev). If an entity body is part of the request (e.g. POST or PUT) it will wait this long between successful read() system calls.

graceful_shutdown(sub { .... })

Causes Feersum to initiate a graceful shutdown of all outstanding connections. No new connections will be accepted. The reference to the socket provided in use_socket() is kept.

The sub parameter is a completion callback. It will be called when all connections have been flushed and closed. This allows one to do something like this:

    my $cv = AE::cv;
    my $death = AE::timer 2.5, 0, sub {
        fail "SHUTDOWN TOOK TOO LONG";
        exit 1;
    Feersum->endjinn->graceful_shutdown(sub {
        pass "all gracefully shut down, supposedly";
        undef $death;

Not really a method so much as a static function. Works similar to EV's/AnyEvent's error handler.

To install a handler:

    no strict 'refs';
    *{'Feersum::DIED'} = sub { warn "nuts $_[0]" };

Will get called for any errors that happen before the request handler callback is called, when the request handler callback throws an exception and potentially for other not-in-a-request-context errors.

It will not get called for read timeouts that occur while waiting for a complete header (and also, until Feersum supports otherwise, time-outs while waiting for a request entity body).

Any exceptions thrown in the handler will generate a warning and not propagated.


Override Feersum's notion of what SERVER_HOST and SERVER_PORT should be.


Compile Time Options

There are a number of constants at the top of Feersum.xs. If you change any of these, be sure to note that in any bug reports.


Defaults to 64. Controls how many headers can be present in an HTTP request.

If a request exceeds this limit, a 400 response is given and the app handler does not run.


Defaults to 128. Controls how long the name of each header can be.

If a request exceeds this limit, a 400 response is given and the app handler does not run.


Defaults to ~2GB. Controls how large the body of a POST/PUT/etc. can be when that request has a Content-Length header.

If a request exceeds this limit, a 413 response is given and the app handler does not run.

See also BUGS



Together, these tune how data is read for a request.

Read buffers start out at READ_INIT_FACTOR * READ_BUFSZ bytes. If another read is needed and the buffer is under READ_BUFSZ bytes then the buffer gets an additional READ_GROW_FACTOR * READ_BUFSZ bytes. The trade-off with the grow factor is memory usage vs. system calls.


Controls how response data is written to sockets. If enabled (the default) the event loop is used to wait until the socket is writable, otherwise a write is performed immediately. In either case, non-blocking writes are used. Using the event loop is "nicer" but perhaps introduces latency, hence this option.


FLASH_SOCKET_POLICY_SUPPORT defaults to disabled.

When it's enabled, Feersum will detect a Flash <policy-file-request/> packet and respond with the FLASH_SOCKET_POLICY string.

The default FLASH_SOCKET_POLICY string looks like this:

    <?xml version="1.0"?>
    <!DOCTYPE cross-domain-policy SYSTEM "/xml/dtds/cross-domain-policy.dtd">
      <site-control permitted-cross-domain-policies="master-only"/>
      <allow-access-from domain="*" to-ports="*" secure="false"/>

Since that's fairly wide-open, you may not wish to enable FLASH_SOCKET_POLICY_SUPPORT.

Note that this feature likely won't work if you use a front-end HTTP server (e.g. nginx) since the request isn't valid HTTP.


Controls the size of the main write-buffer structure in Feersum. Making this value lower will use slightly less memory per connection at the cost of speed (and vice-versa for raising the value). The effect is most noticeable when you're app is making a lot of sparce writes. The default of 64 generally keeps usage under 4k per connection on full 64-bit platforms when you take into account the other connection and request structures.

NOTE: FEERSUM_IOMATRIX_SIZE cannot exceed your OS's defined IOV_MAX or UIO_MAXIOV constant. Solaris defines IOV_MAX to be 16, making it the default on that platform. Linux and OSX seem to set this at 1024.


For non-threaded perls >= 5.12.0, this defaults to enabled.

When enabled, Feersum will "steal" the contents of temporary lexical scalars used for response bodies. The scalars become undef as a result, but due to them being temps they likely aren't used again anyway. Stealing saves the time and memory needed to make a copy of that scalar, resulting in a mild to moderate performance boost.

This egregious hack only extends to non-magical, string, PADTMP scalars.

If it breaks for your new version of perl, please send stash a note (or a pull request!) on github.

Worth noting is that a similar zero-copy effect can be achieved by using the psgix.body.scalar_refs feature.


Please report bugs using

Keep-alive is ignored completely.

Currently there's no way to limit the request entity length of a streaming POST/PUT/etc. This could lead to a DoS attack on a Feersum server. Suggested remedy is to only run Feersum behind some other web server and to use that to limit the entity size.

Although not explicitly a bug, the following may cause undesirable behavior. Feersum will have set SIGPIPE to be ignored by the time your handler gets called. If your handler needs to detect SIGPIPE, be sure to do a local $SIG{PIPE} = ... (perlipc) to make it active just during the necessary scope.


Feersum Git: git://

picohttpparser Git: git://


Jeremy Stashewsky,


Tatsuhiko Miyagawa for PSGI and Plack.

Marc Lehmann for EV and AnyEvent (not to mention JSON::XS and Coro).

Kazuho Oku for picohttpparser.

Luke Closs (lukec), Scott McWhirter (konobi), socialtexters and for initial feedback and ideas. Audrey Tang and Graham Termarsch for XS advice.

Hans Dieter Pearcey (confound) for docs and packaging guidance.

For bug reports: Chia-liang Kao (clkao), Lee Aylward (leedo)

Audrey Tang (au) for flash socket policy support.


Copyright (C) 2011 by Jeremy Stashewsky

Portions Copyright (C) 2010 Socialtext Inc.

This library is free software; you can redistribute it and/or modify it under the same terms as Perl itself, either Perl version 5.8.7 or, at your option, any later version of Perl 5 you may have available.

picohttpparser is Copyright 2009 Kazuho Oku. It is released under the same terms as Perl itself.