There seems to be some concern amongst users about the massive surgery we did on libplasma this past month. The concern stems from the idea that these changes will work against the stabilization of libplasma and result in prolonging a "beta" quality to plasma itself.
It's important to first understand that these changes were planned, even before 4.0. We knew that "widgets on canvas" was coming and so we could eventually remove our own layouting and QWidget bridges at some point for something that was more robust and less of a hack. We also knew that being a first revision of a library API (application programmer's interface) that would see a lot of usage, it was highly likely that changes would come to be needed or wanted. It's nearly impossible to foretell exactly what will be used and required in such an API, and it's really unrealistic to hope that the first draft of the semantics in an API will be optimal any more than it is to expect the first draft of a novel to not need any editing and revising.
So before 4.0 came out I told everyone that libplasma would not be binary compatible between 4.0 and 4.1 so that we could reshape the API as needed to make it last longer. I told everyone that we'd be porting to widgets on canvas when we could begin using Qt 4.4. I told everyone that we'd be replacing the icons-on-desktop implementation with something more robust that offered access to the same features.
This was all known and planned for before 4.0 was released. None of them could be done in 4.0 for various reasons (such as Qt 4.4 not being available to us), and so we lived with various hacks and bodges.
Moving to the development of 4.1, we executed on these changes. Right now plasma is approximately as stable and in most areas a bit more featureful than what is in 4.0.4, even with all those backports we did to catch up the 4.0 branch. There are a few regressions that remain in the development tree right now, but they are disappearing with rapidity. But being able to finally see things happen that we always imagined, such as the device notifier expanding from an icon to the full view when dropped onto the desktop (aka "adjusting the visualization in response to the form factor"), is very rewarding to watch happen.
There has been one downside to the massive changes: it took time away from creating more new features. So some of the plasmoids I wanted to have working for 4.1 may not happen and get punted to 4.2 instead. Thankfully with all this work done, features that do get worked on from here out should be easier to accomplish than when working with the 4.0 API.
Thankfully we will not have to repeat this process again during the 4.x time frame. We will be able to move forward with keeping the API in place. We will add to it as needed, which isn't disruptive, while modifications such as these ones for 4.1 simply won't be on the table.
So if you are concerned about the recent changes made impacting stability, I thank you for your concern and empathize with how one could arrive at such a conclusion. Thankfully, these changes have been made specifically to answer your concerns, as well as mine, about stability and feature completeness both in 4.1 and beyond.
I hope that clears up a few things. If not, you may wish to track the betas and release candidates as they begin to appear near the end of May and see for yourself.