blob: 7d240c8e77fc23c7190eada6c0f8f4715a2e7305 [file] [log] [blame]
<?xml version="1.0"?>
<document>
<properties>
<author email="ceki@apache.org">Ceki Gulcu</author>
<title>Release plan for log4j 1.3</title>
</properties>
<body>
<section name="Plan? What plan?">
<p>The following paragragh was adapted from the <a
href="http://jakarta.apache.org/cactus/todo.html">Cactus
project's roadmap</a> with the author's permission.
</p>
<p>Our users keep inventing better ways and adding new
requirements. The downside is that our todo list keeps
growing. The upside is that there is plenty of work to go
around. If you are interested in participating, send an email
on the log4j-dev@ mailing list stating your interest. You'll
be promptly enrolled. We're always looking for help! Don't be
put off if in the "Volunteer" column already has a person
listed. Programming is fun, especially if it is done in a
team.
</p>
<p>
</p>
</section>
<section name="Workplan for log4j 1.3">
<p>The workplan for log4j 1.3 is not final. It is included here
to give you an idea of the future. The items are not listed in
any particular order. As always, there is no scheduled
release date. The lack of schedule suprises and disturbs some
people. Writing good software, like good cooking, takes
time. If we make you wait, it is to create a better and more
reliable product.
</p>
<table border="1" cellpadding="3" cellspacing="2">
<tr>
<th>Label</th>
<th>Comment</th>
<th>Volunteer</th>
<th>Status</th>
</tr>
<tr bgcolor="DDDDDD">
<td><b>test cases</b></td>
<td>
<p>Writing test cases is not the most sexy part of
software development but it is one of the most
important. Automated test cases allow us to catch bugs
as early is possible. It is strongly recommended to add
a new test case with each new feature or component.</p>
<p>Existing <em>Perl</em> language based test cases are
gradually being migrated to junit (all-Java)based test
cases. The new tests are placed under the
<code>tests/</code> directory. It should be thus
possible for participants in the project to understand
the stucture of our tests and add tests for their
components.
</p>
</td>
<td>All committers</td>
<td>ongoing effort</td>
</tr>
<tr>
<td>Extensible XML configuration files</td>
<td>
<p>The DOMConfigurator is complex and not very
flexible. It can only deal with elements that the
developer knew about at compilation time. This has been
an important drawback in the design of several appenders
such as the the SMTPAppender and the
RollingFileAppenders and its variants. These appenders
need to delegate certain task to sub-components which
are configured separately.
</p>
<p>The support for extensible configuration files will be
based on the code in commons-digester authored by Craig
McClanahan and Scott Sanders. They graciously granted
permission for the modification of their code and its
inclusion in log4j.</p>
</td>
<td>Ceki</td>
<td>design board</td>
</tr>
<tr>
<td>Log4j Domains</td>
<td>
<p>This is a very powerful and innovative concept that
extends the notion of hierarchical loggers. It will also
allow dynamic logging with pin-point accuracy. It was
first suggested to me by Scott Stark of <a
href="http://www.jboss.org">JBoss</a> fame.
</p>
</td>
<td>Ceki</td>
<td>design board</td>
</tr>
<tr>
<td>Multiple implementations of Logger</td>
<td>
<p>Based on <code>RepositorySelectors</code> introduced in
log4j 1.2, the user will be able to replace the
<code>Logger</code> implementation. Several
implementations will be provided offering different
properties and functionality although none of the
implementations will add new public methods.
</p>
</td>
<td>?</td>
<td>vaporware</td>
</tr>
<tr>
<td>Improvements to Chainsaw</td>
<td>
<p>Visualisation and dynamic filtering of logging event is
bound to be a very important area of activity in the
future. A number of important improvements to chainsaw
have been suggested.
</p>
</td>
<td>Oliver Burn</td>
<td>under discusison</td>
</tr>
<tr>
<td>Custom conversion characters in PatternLayout</td>
<td>Users often want to add new conversions characters or
override the existing ones. This should be made easy using
new configuration directives. This feature would use the
extensions to XML configuration language mentioned
above.</td>
<td>?</td>
<td>not started</td>
</tr>
<tr>
<td>Overture to other programming languages</td>
<td>It is higly desriable to allow log4j ports in other languages
to access log4j services in a language independent way.
</td>
<td>?</td>
<td>not started</td>
</tr>
<tr>
<td>Strategy based rollovers</td>
<td>
<p>Contrary to our own DailyRollingFileAppender, Avalon's
logkit has a nice and clean implementation for rolling
files. See the
<code>org.apache.log.output.io.rotate</code> package for
exact details.
</p>
<p>Their implementation is based on strategies which are
sub-components of appender. We will be able to configure
such sub-components with the new XML configuration
scripts.
</p>
</td>
<td>Kevin Steppe</td>
<td>not started</td>
</tr>
<tr>
<td>Redesign of configure and watch architecture in
configurators</td>
<td>This is a very useful feature and the current architecture is not very good.
<p>Contributions have been received by Mark Womack and others.</p>
<p>See
<pre>
http://www.mail-archive.com/log4j-dev@jakarta.apache.org/msg01390.html
http://www.mail-archive.com/log4j-user@jakarta.apache.org/msg00666.html
http://marc.theaimsgroup.com/?t=101010070500002&amp;r=1&amp;w=2
</pre>
</p>
</td>
<td>Mark Womack</td>
<td>initial implementation</td>
</tr>
<tr>
<td>Performance improvements to LoggingEvent serialization</td>
<td>
<p>Ole Dalgaard has shown that by implementing the
java.io.Externalizable interface instead of
java.io.Serializable in the LoggingEvent class, the
speed of serialiazation is increased by a factor of 4 or
more.
</p>
</td>
<td>Ole Dalgaard?</td>
<td>initial implementation</td>
</tr>
</table>
</section>
<section name="Workplan for log4j 1.2">
<p>
Here is workplan for the 1.2 release.
</p>
<table border="1" cellpadding="3" cellspacing="2">
<tr>
<th>Label</th>
<th>Comment</th>
<th>Volunteer</th>
<th>Status</th>
</tr>
<tr bgcolor="DDDDDD">
<td>More test cases</td>
<td>We need more automated test cases to catch bugs as early
is possible.It is highly recommended to add a new test case
with each new feature or component.</td>
<td>All committers</td>
<td>ongoing effort</td>
</tr>
<tr>
<td>Removal of deprecated methods</td>
<td>Deprecated <code>setOption</code> and <code>getOption</code>
methods in appender and layouts should be removed.</td>
<td>Ceki</td>
<td>done</td>
</tr>
<tr bgcolor="DDDDDD">
<td>JDBCAppender</td>
<td>
<p>We currently have two competing JDBCAppenders: one by
Thomas Fenner and the other by Kevin Steppe.</p>
<p>Kevin Steppe stood up and did it.</p>
</td>
<td>Kevin Steppe</td>
<td>done</td>
</tr>
<tr>
<td>Log4j in applets</td>
<td>In order to minimize network traffic, the size of log4j-core.jar
needs to be reduced to at most 50KB.
<p>Log4jME has been released. It's less than 20KB.</p>
</td>
<td>Ceki</td>
<td>done</td>
</tr>
<tr bgcolor="DDDDDD">
<td>Improved documentation</td>
<td>Log4j documentation needs to be enhanced with configuration
examples and much more hand-holding.</td>
<td>Ceki</td>
<td>ongoing effort</td>
</tr>
<tr>
<td valign="top">Mapped Diagnostic Contexts</td>
<td>Mapped Diagnostic Contexts are similar to the NDC
except that the MDC is a string to string map instead of
a stack that the user sets when entering a special
context. The <code>PatternLayout</code> has to be
enhanced to support this by extending the %x pattern to
accept an argument. Here is an example:
<pre>
ConversionPattern=3D%d %p %c %x{server} %x{host} - %m%n
</pre>
User code:
<pre>
{
MDC.put("server", "totoServer");
MDC.put("host", "someHost");
logger.debug("Hello");
}
</pre>
will print:
<pre>2000-01-24 10:00:00,000 DEBUG totoServer someHost - Hello</pre>
<p>To make MDCs truly user friendly
<code>ThreadLocal</code> variables are required. This
will allow the MDC to be inherited by child
threads. <code>ThreadLocal</code> are only supported
under JDK 1.2 and above. In JDK 1.1, the MDC will not
work but won't harm the user application either.</p>
</td>
<td>Ceki</td>
<td>done</td>
</tr>
<tr bgcolor="DDDDDD">
<td>Enhanced variable substitution support in DOMConfigurator</td>
<td></td>
<td>Ceki</td>
<td>done</td>
</tr>
<tr>
<td>FallbackErrorHandler</td>
<td>The FallbackErrorHandler implements the ErrorHandler
interface such that a secondary appender may be
specified. This secondary appender takes over if the primary
appender fails for whatever reason.
<p>The DOMConfigurator needs to be extended to support the
FallbackErrorHandler</p>
</td>
<td>Ceki</td>
<td>implemented, requires further testing</td>
</tr>
<tr>
<td>Ensure backward compatibility of LoggingEvent
objects</td>
<td>To avoid deployment problems we must ensure that
LoggingEvent objects are compatible between 1.2 and 1.1.3.
<p>Robert Bushman has proposed a very innovative way for
solving this problem. See <a
href="http://www.mail-archive.com/log4j-dev@jakarta.apache.org/msg01397.html">
http://www.mail-archive.com/log4j-dev@jakarta.apache.org/msg01397.html</a>
for further details.
</p>
</td>
<td>Ceki</td>
<td>implemented, manually tested, requires automated test cases</td>
</tr>
</table>
</section>
</body>
</document>