Tapestry Component Library Auto-Discovery
After working on application component
auto-discovery (see earlier blog postings) I wanted to get Tapestry to
auto-discover Tapestry component
libraries.It wasn't difficult and I
got my application specification down to:
<description><![CDATA[ A Tapestry Test Application ]]></description>
<property name="org.apache.tapestry.disable-caching"
value="true"/>
<property name="org.apache.tapestry.visit-class"
value="com.mjhenderson.users.tapestry.Visit"/>
<service name="users.membership"
class="com.mjhenderson.users.tapestry.engine.MembershipEngineService"/>
</application>
The two component library references are
gone, auto-discovery adds them to the application
specification.Auto-discovery works by
resolving all files with the name
tapestry-library.properties
on the class path and reading each on in to a
java.util.Properties. The file needs 2 property settings in
it:
tapestry.library.id=springinspector
tapestry.library.specification-path=/com/mjhenderson/tapestry/springinspector/springinspector.library
I modified the
org.apache.tapestry.ApplicationServlet class to perform library auto-discovery
after reading the application specification
file:
private void addAutoDiscoveredLibrarySpecifications(IApplicationSpecification appSpec) {
LibraryLocator locator = new LibraryLocator();
Iterator libraries = locator.getAutoDiscoveredLibraries().iterator();
while (libraries.hasNext()) {
LibraryLocator.LibraryInfo info = (LibraryLocator.LibraryInfo)libraries.next();
if (appSpec.getLibrarySpecificationPath(info.id) == null) {
appSpec.setLibrarySpecificationPath(info.id, info.path);
}
}
}
LibraryLocator does the
work:
public List getAutoDiscoveredLibraries() {
List list = new ArrayList();
try {
Enumeration resources = Thread.currentThread().getContextClassLoader().getResources("tapestry-library.properties");
while (resources.hasMoreElements()) {
URL url = (URL)resources.nextElement();
Properties p = readProperties(url);
String id = p.getProperty("tapestry.library.id");
String path = p.getProperty("tapestry-library.specification-path");
if (id != null && path != null) {
LibraryInfo info = new LibraryInfo();
info.id = id;
info.path = path;
list.add(info);
}
}
}
catch (IOException ex) {
ex.printStackTrace();
}
return list;
}
I expect that the HiveMind integration in
Tapestry 3.1 will make this type of auto-discovery of libraries and components a
standard feature with a much more elegant implementation. I hope so, reducing
configuration requirements to a bare minimum, with reasonable defaults can lower
the barrier to entry for developers wanting to take Tapestry out for a spin.
This lack of configuration requirements could even be a distinguishing factor of
Tapestry.Until then I'll use a patched
tapestry-3.0.jar
and a patched tapestry-contrib-3.0.jar or apply a
source
patch to Tapestry.
Posted: Mon - August 16, 2004 at 01:34 PM
|
