New features

Unzipping of zipped P2 repositories

It is now possible to add the "P2Unzipper" aspect to a channel, which will extract plugins and features from a zipped P2 repository. Categories are not extracted and binaries additions are skipped as well. This will change in the future.

Building up categories based on `category.xml`

Uploading an Eclipse category.xml file using the P2 Category Generator now allows one to re-use these categories files and let Package Drone build up P2 categories as defined in the file, using the Eclipse PDE toolchain.

Collapsible tree view

The tree view has been enhanced to allow expanding/collapsing of child element, and also show some information about the type of an artifact (classifier).

Channel aspect dependencies

Channel aspects now can declare which other aspects they require. These can be added in one step to the channel. This should help everyone not reading the text (like myself 😉 ) to not forget adding required aspects. Also can the description now contain HTML text.

Binary channel level cache

There is a persistent cache feature on the channel level. This can be used by channel aggregators to store output for later use. For example the P2 repository aspect creates the full meta data information during the aggregation run, stores it in the cache so that the P2 adapter later can re-use this information for serving the P2 repository and does not need to build up the channel view again. This bring a great improvement of the loading speeds for the P2 adapter.

The reason that this information was not stored as artifact itself simply is that this would change the channel again, so one aggregation run would trigger the next one. Also can cache entries be referenced by ID and therefore loaded much quicker.

Changes

The default view for artifacts is the tree view now
All aspects create a classifier information now. This will only be visible if the channel aspects get refreshed at least once.

Improvements and fixed issues

There were a few locations where the active flag of UI elements was not rendered correctly
The loading of artifact information with meta data has been improved
There is a "classifier" column in the tree view which show which type of artifacts this is
Byte sizes are now also reported as KiB, MiB and GiB
The job view now has a progress meter

Known bugs and issues

Although Package Drone uses streaming API to send data over to the database, MySQL seems to simply implement the JDBC streaming API by allocating huge memory arrays. This can result in a OutOfMemory exceptions. Although this is not an issue of Package Drone, but the MySQL JDBC driver, it is a problem since MySQL currently is the only supported database. Hopefully this will change in the future (@CptMauli).

So for know, the solution is to increase the amount of memory assigned to Package Drone (using good old -Xmx).