featherlite V3.3 released
During the last few month we spent a lot of work on the featherlite framework. This is mainly through many new requirements which arose through discussions with our customers. This led to numerous improvements in performance, integration and debugging.
Since fast rescheduling of workflows is one of the requirements of a customer implementation we support, we took care of improving the performance. To improve the performance further we moved the client updates of objects changed from the services to the underlying commands and now collect all objects to be updated in the in-memory transaction. This completes our long term work on the in-memory transactions we use in our framework. Now objects touched in one transaction are guaranteed to be published to observers and clients only once, reducing the time loss due to multiple object updates in a single transaction. And it simplifies the work to be done to implement new project specific services, since the programmer no longer needs to take care of object updates, since they are now delegated to the commands called from the services.
Since featherlite applications in general are embedded in an existing landscape, we improved the integration part of the framework. We added a new connection view which gives the user a better insight on which connections are registered in the connection handler and what their state is. In addition we added new actions, services and commands to manipulate the connection properties and state. Further we added new demonstration scenarios implementing a text file, an XML via TCP/IP and an Excel interface using Apache POI.
For this release we at long last implemented a logging console which remotely listens to an RMI appender we implemented on top of the log4j framework as well as well as a log file viewer to load the log files from a remote featherlite server. The console has all the features we are used to from the eclipse console and the log file viewer supports regular expression searches and filtering. With this the trouble shooting using the log4j files gets much more comfortable than the work with the operating system file viewer we had to use so far.
Last but not least we improved the Gantt chart client to reach a better usability, as the featherlite Gantt chart is moving more and more from its shadow existence as a debugging tool for the developer to an end user client tool. Row and column resizing via mouse, alpha blending for the task items and opening of value rows for selected resources are only some of the new features we added to the Gantt chart view. Since the new features are better shown than explained, we are currently preparing a screencast showing all the features. With these new features we feel that the Gantt chart is moving toward a rich Gantt chart client tool.
