The externalizer uses the name org.apache.tapestry.AssetExternalizer.application name
. It configures itself using two additional
properties (searching in
getPropertySource() .
| Parameter | Description |
|---|---|
org.apache.tapestry.asset.dir |
The directory to which assets will be copied. |
org.apache.tapestry.asset.URL |
The corresponding URL for the asset directory. |
If either of these parameters is null, then no externalization occurs. Private assets will still be available, just less efficiently, as the application will be invoked via its servlet and, ultimately, the AssetService will need to retrieve the asset.
Assets maintain thier directory structure when copied. For example,
an asset with a resource path of /com/skunkworx/Banner.gif would
be copied to the file system as dir/com/skunkworx/Banner.gif and
would have a URL of URL/com/skunkworx/Banner.gif.
The externalizer will create any directories as needed.
The externalizer will not overwrite existing files. When a new version of the application is deployed with changed assets, there are two deployment stategies:
When using the second approach, it is best to use a directory that has
a version number in it, for example, D:/inetpub/assets/0 mapped to the URL
/assets/0. When a new version of the application is deployed, the trailing
version number is incremented from 0 to 1.
| Constructor Summary |
|---|
|
No description provided. |
| Method Summary | |
|---|---|
| protected void |
No description provided. |
| static AssetExternalizer |
Gets the externalizer singleton for the application. |
| String |
Gets the URL to a private resource. |
| Methods inherited from java.langObject |
|---|
Each Tapestry application within a single ServletContext will have its own externalizer; they are differentiated by the application name.
If the asset directory and URL are not configured, then returns null.
Otherwise, the asset is copied out to the asset directory, the URL is constructed (and recorded for later) and the URL is returned.
This method is not explicitly synchronized but should work
multi-threaded. It synchronizes on the internal
Map used to map resource paths to URLs.