NAME
    ElasticSearch::Transport::AECurl - AnyEvent::Multi::Curl (libcurl)
    backend for ElasticSearch

VERSION
    version 0.03

SYNOPSIS
        use ElasticSearch;
        my $e = ElasticSearch->new(
            servers         => 'search.foo.com:9200',
            transport       => 'aecurl',

            timeout         => 30,
            max_concurrency => 0,
            proxy           => 'foo.bar.com',
            ip_resolve      => 4|6|undef,
        );

        # blocking request
        $e->cluster_health->recv;

        # non-blocking request
        $e->cluster_health->cb( sub {
            if ($@) {
                log "An error occurred: $@";
            } else {
                my $response = shift;
                log $response;
            }
        });
        AE::cv->recv();

        # fire-and-forget vs scoped
        {
            $e->delete_index(index => 'foo');
            $e->delete_index(index => 'bar')->cb( sub { print "Done"});
            my $cv = $e->delete_index(index=>'baz');
        }
        AE::cv->recv;
        #   - foo and bar will be deleted
        #   - baz will not be deleted

DESCRIPTION
    ElasticSearch::Transport::AECurl uses AnyEvent::Multi::Curl (and thus
    "libcurl") to talk to ElasticSearch asynchronously over HTTP.

USING AECurl
    Any request to ElasticSearch returns an AnyEvent::CondVar. You have
    three options for how you use them:

  Blocking
        $cv = $e->cluster_health;
        $result = $cv->recv;

    When you call "recv()" on a CondVar, your program will block until that
    CondVar is ready to return a value.

    If an error was thrown, then "recv()" will "die". You will need to wrap
    "recv()" in "eval" if you don't want to die.

    If your $cv goes out of scope, then the request will be aborted.

  Callback
        $e->cluster_health->cb( sub {
            if ($@) {
                log "Error $@";
            }
            else {
                my $result = shift;
                log "$result"
            }
        })
        AE::cv->recv()

    If you set a callback on a CondVar, then the callback will be called
    once the CondVar is ready (which will only happen after you start the
    event loop).

    In the callback, $@ will contain any error, otherwise the result (if
    any) will be the first value in @_.

    Once you set a callback on a CondVar, it will not be aborted when it
    goes out of scope.

  Fire-and-Forget
        $e->delete_index(index=>'foo');

    If a request is called in "void" context, then it will be executed once
    the event loop is started. No errors will be thrown, even if the request
    does not complete succesfully.

    It will not be aborted with a change in scope, because there is no
    scope. If you exit the application without running an event loop, then
    any pending requests will not be run.

BLOCKING METHODS
    "scrolled_search()" in ElasticSearch and "reindex()" in ElasticSearch
    will be executed synchronously.

SETTINGS
    See AnyEvent::Curl::Multi for an explanation of the "max_concurrency()",
    "max_redirects()", "proxy()" and "ip_resolve()"

SEE ALSO
    *   ElasticSearch

    *   ElasticSearch::Transport

    *   ElasticSearch::Transport::HTTPLite

    *   ElasticSearch::Transport::HTTPTiny

    *   ElasticSearch::Transport::Curl

    *   ElasticSearch::Transport::AEHTTP

    *   ElasticSearch::Transport::Thrift

    1;

AUTHOR
    Clinton Gormley <drtech@cpan.org>

COPYRIGHT AND LICENSE
    This software is copyright (c) 2012 by Clinton Gormley.

    This is free software; you can redistribute it and/or modify it under
    the same terms as the Perl 5 programming language system itself.