Oliver,
Do you mean that this is almost independent of the rest of Slide, or is it still dependent on the large part of the Slide infrastructure?
Carlos
Oliver Zeigermann wrote:
Folks,
if things work out the way I plan I will add the WebDAV Construction Kit to the proposals section of the Slide CVS HEAD today.
It contains classes that help you to make your proprietary system WebDAV aware without much knowledge of either Slide and then also WebDAV to an extend that it should work with the usual Windows, Mac and Linux clients.
It emerged out of the experience that Slide's storing system is most flexible and generic, but too complex for many people who just want to add the WebDAV interface to their systems. Additionally, it contains a JAAS login module that allows you to bypass Slide's user managerment and the complete internal access checking. All this is done by your proprietary system. This, however, is optional to you.
All this is the largest hack I have ever done as it relies on many implementation details and sequences of calls - at least it should be the standard hack that will be maintained and adapted to Slide as we go along. It is important to understand that all this works only when you access Slide through its WebDAV interface and only if you restrict yourself to the core WebDAV methods. It is likely not to work when you use the native Slide interface. It will not work with ACL, BIND, SEARCH and DeltaV methods.
The main interfaces -------------------
To make Slide work as your WebDAV layer you will have to implement a single interface: org.apache.slide.simple.store.BasicWebdavStore. It contains some call back methods which do not use any Slide classes except for the exceptions, which I thought are pretty obvious and thus it would have been silly to more or less clone them. Here they are for a short reference - the complete ones I will check in will have much more detailed documentation:
- begin, commit, rollback: begins, commits, aborts a request / transaction
- objectExists, isFolder, isResource: existence checks, collection vs. content checks
- createFolder, createResource: creation of a collection or a content resource
- removeObject: deletion of a collection or a content resource
- setResourceContent, getResourceContent: storing and retrieval of a content resource
- getLastModified, getCreationDate, getResourceLength: retrieval of basic meta data
- getChildrenNames: retrieval of the children of a collection
This is all Slide and WCK need to make basic WebDAV operations work. There are optional extension interfaces that add full property access and locking if your system supports them.
The second main interface is even simpler and allows you to login into Slide using your proprietary user database. The glue code will pass session and connection information to your store as well. All this allows to effectively bypass Slide's security code and leave this to your system. All you have to implement are three methods of the org.apache.slide.simple.authentication.SessionAuthenticationManager.
- getAuthenticationSession (2x one with user/password one with user only): asks if a certain user exists and authenticated itseld correctly and if so an object that allows access to your system, i.e. a session or connection or anything similiar, is passed back; in case the user has already authenticated correctly before only the user with no password is passed
- closeAuthenticationSession: closes the session / terminates the connection to your system
Using this interface the glue code finds out if a user exists and authenticated itself correctly and passes all relevant information to the basic WebDAV store for you to make authenticated requests to your system later. All this is optional, you still can use Slide's internal security with your WebDAV store if this is not appropriate for you. For your convenience there also is a connection pool available.
Reference Implementation ------------------------
For the store, the session manager and the pooling connection manager there are reference implementations to the file system to help you get going. However, the ones for sessions and connection management are dummies only as the file system neither needs sessions nor connections.
The reference implementation of the store part and also parts of the glue code are heavily inspired by the work of Alon who provided the simple file store to us - so credit goes to Alon. The new implementation of the file store extends Alon's in adding arbitrary properties (optional) and locks (optional) and proper handling of access violations.
Building --------
WCK works with both the current versions of Slide 2.1 and the CVS HEAD and comes with a build script that creates a full working web archive to be deployed in Tomcat. Which one it will compile to and package with is to be determined in build.properties. If this isn't obvious details will follow upon request. When WCK matures it might be possible to even release it as a distribution of its own, but this is something to discuss.
Bottom Line -----------
What I need from you is feedback as detailed as possible if this suits your needs our what should be changed resp. added to make this what you need.
Additionally, I am pretty sure I have forgotten the most important parts, so remind me and ask for them, please...
Finally, have in mind this is work in progress so changes to the API are somewhat likely.
Have fun checking this out and cheers
Oliver
--------------------------------------------------------------------- To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED]
--------------------------------------------------------------------- To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED]
