<?xml version="1.0" encoding="UTF-8" ?>
<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Strict//EN" "http://www.w3.org/TR/xhtml1/DTD/xhtml1-strict.dtd">
<html xmlns="http://www.w3.org/1999/xhtml" lang="en">
<head>
<meta http-equiv="Content-Type" content="text/html; charset=UTF-8" />
<link rel="stylesheet" href=".resources/doc.css" charset="UTF-8" type="text/css" />
<link rel="shortcut icon" href=".resources/report.gif" type="image/gif" />
<title>JaCoCo - Java Agent</title>
</head>
<body>
<div class="breadcrumb">
<a href="../index.html" class="el_report">JaCoCo</a> >
<a href="index.html" class="el_group">Documentation</a> >
<span class="el_source">Java Agent</span>
</div>
<div id="content">
<h1>Java Agent</h1>
<p>
JaCoCo uses class file instrumentation to record execution coverage data.
Class files are instrumented on-the-fly using a so called Java agent. This
mechanism allows in-memory pre-processing of all class files during class
loading independent of the application framework.
</p>
<p class="hint">
If you use the <a href="ant.html">JaCoCo Ant tasks</a> or
<a href="maven.html">JaCoCo Maven plug-in</a> you don't have to care about the
agent and its options directly. This is transparently handled by the them.
</p>
<p>
The JaCoCo agent collects execution information and dumps it on request or
when the JVM exits. There are three different modes for execution data ouput:
</p>
<ul>
<li>File System: At JVM termination execution data is written to a local
file.</li>
<li>TCP Socket Server: External tools can connect to the JVM and retrieve
execution data over the socket connection. Optional execution data reset
and execution data dump on VM exit is possible.</li>
<li>TCP Socket Client: At startup the JaCoCo agent connects to a given TCP
endpoint. Execution data is written to the socket connection on request.
Optional execution data reset and execution data dump on VM exit is
possible.</li>
</ul>
<p>
The agent <code>jacocoagent.jar</code> is part of the JaCoCo distribution and
includes all required dependencies. A Java agent can be activated with the
following JVM option:
</p>
<pre>
-javaagent:<i>[yourpath/]</i>jacocoagent.jar=<i>[option1]</i>=<i>[value1]</i>,<i>[option2]</i>=<i>[value2]</i>
</pre>
<p>
The JaCoCo agent accepts the following options:
</p>
<table class="coverage">
<thead>
<tr>
<td>Option</td>
<td>Description</td>
<td>Default</td>
</tr>
</thead>
<tbody>
<tr>
<td><code>destfile</code></td>
<td>Path to the output file for execution data.</td>
<td><code>jacoco.exec</code></td>
</tr>
<tr>
<td><code>append</code></td>
<td>If set to <code>true</code> and the execution data file already
exists, coverage data is appended to the existing file. If set to
<code>false</code>, an existing execution data file will be replaced.
</td>
<td><code>true</code></td>
</tr>
<tr>
<td><code>includes</code></td>
<td>A list of class names that should be included in execution analysis.
The list entries are separated by a colon (<code>:</code>) and
may use wildcard characters (<code>*</code> and <code>?</code>).
Except for performance optimization or technical corner cases this
option is normally not required.
</td>
<td><code>*</code> (all classes)</td>
</tr>
<tr>
<td><code>excludes</code></td>
<td>A list of class names that should be excluded from execution analysis.
The list entries are separated by a colon (<code>:</code>) and
may use wildcard characters (<code>*</code> and <code>?</code>).
Except for performance optimization or technical corner cases this
option is normally not required.
</td>
<td><i>empty</i> (no excluded classes)</td>
</tr>
<tr>
<td><code>exclclassloader</code></td>
<td>A list of class loader names that should be excluded from execution
analysis. The list entries are separated by a colon
(<code>:</code>) and may use wildcard characters (<code>*</code> and
<code>?</code>). This option might be required in case of special
frameworks that conflict with JaCoCo code instrumentation, in
particular class loaders that do not have access to the Java runtime
classes.
</td>
<td><code>sun.reflect.DelegatingClassLoader</code></td>
</tr>
<tr>
<td><code>inclbootstrapclasses</code></td>
<td>Specifies whether also classes from the bootstrap classloader should
be instrumented. Use this feature with caution, it needs heavy
includes/excludes tuning.
</td>
<td><code>false</code></td>
</tr>
<tr>
<td><code>sessionid</code></td>
<td>A session identifier that is written with the execution data. Without
this parameter a random identifier is created by the agent.
</td>
<td><i>auto-generated</i></td>
</tr>
<tr>
<td><code>dumponexit</code></td>
<td>If set to <code>true</code> coverage data will be written on VM
shutdown. The dump can only be written if either <code>file</code> is
specified or the output is <code>tcpserver</code>/<code>tcpclient</code>
and a connection is open at the time when the VM terminates.
</td>
<td><code>true</code></td>
</tr>
<tr>
<td><code>output</code></td>
<td>Output method to use for writing coverage data. Valid options are:
<ul>
<li><code>file</code>: At VM termination execution data is written to
the file specified in the <code>destfile</code> attribute.</li>
<li><code>tcpserver</code>: The agent listens for incoming connections
on the TCP port specified by the <code>address</code> and
<code>port</code> attribute. Execution data is written to this
TCP connection.</li>
<li><code>tcpclient</code>: At startup the agent connects to the TCP
port specified by the <code>address</code> and <code>port</code>
attribute. Execution data is written to this TCP connection.</li>
<li><code>none</code>: Do not produce any output.</li>
</ul>
Please see the security considerations below.
</td>
<td><code>file</code></td>
</tr>
<tr>
<td><code>address</code></td>
<td>IP address or hostname to bind to when the output method is
<code>tcpserver</code> or connect to when the output method is
<code>tcpclient</code>. In <code>tcpserver</code> mode the value
"<code>*</code>" causes the agent to accept connections on any local
address.
</td>
<td><i>loopback interface</i></td>
</tr>
<tr>
<td><code>port</code></td>
<td>Port to bind to when the output method is <code>tcpserver</code> or
connect to when the output method is <code>tcpclient</code>. In
<code>tcpserver</code> mode the port must be available, which means
that if multiple JaCoCo agents should run on the same machine,
different ports have to be specified.
</td>
<td><code>6300</code></td>
</tr>
<tr>
<td><code>classdumpdir</code></td>
<td>Location relative to the working directory where all class files seen
by the agent are dumped to. This can be useful for debugging purposes
or in case of dynamically created classes for example when scripting
engines are used.
</td>
<td><i>no dumps</i></td>
</tr>
<tr>
<td><code>jmx</code></td>
<td>If set to <code>true</code> the agent exposes
<a href="./api/org/jacoco/agent/rt/IAgent.html">functionality</a> via
JMX under the name <code>org.jacoco:type=Runtime</code>. Please see
the security considerations below.
</td>
<td><code>false</code></td>
</tr>
</tbody>
</table>
<h2>Security Consideration for Remote Agent Control</h2>
<p>
The ports and connections opened in <code>tcpserver</code> and
<code>tcpclient</code> mode and the JMX interface do not provide any
authentication mechanism. If you run JaCoCo on production systems make sure
that no untrusted sources have access to the TCP server port, or JaCoCo TCP
clients only connect to trusted targets. Otherwise internal information of the
application might be revealed or DOS attacks are possible.
</p>
</div>
<div class="footer">
<span class="right"><a href="@jacoco.home.url@">JaCoCo</a> @qualified.bundle.version@</span>
<a href="license.html">Copyright</a> © @copyright.years@ Mountainminds GmbH & Co. KG and Contributors
</div>
</body>
</html>