EMC’s Innovation Conference is over so I can start to devote a little more time to writing about virtual worlds in the enterprise. For the past few weeks I have been working on my submission for a proof-of-concept for a support center for an in-house application. The effort has largely been one of debugging aspects of realXtend as it goes through an amazing amount of change. If you’ve been following the story of realXtend on the Ugotrade blog or here, you’ll know that realXtend forked from Opensim and periodically resynchronizes with it. The rate of change of both platforms has made it hard to merge the realXtend features back into Opensim easily.
Here are some statistics for both servers over the period August 21 to October 23 illustrating the amount of change in the platforms. Both development teams describe their code as in the alpha stage, and both groups note that people are doing interesting things with the code nonetheless.
|Platform||Total Commits||Total File Changes|
As Tish Shute noted recently, Adam Frisby of DeepThink has finally demonstrated a new plug-in architecture for OpenSim that enables the integration of realXtend features like mesh avatars as optional features. Jani Pirkola, program manager for realXtend, says that he expects his team to switch over to the plug-in architecture in December.
With that as background, it’s easy to understand how one might spend as much time on debugging as on content development. In fact, I had to deal with a major release (0.31), a move of my server to a different subnet and DNS server, two weeks of downtime during the move because of some networking components missing in the new data center, plus frequent attempts to rebuild the server code to incorporate bugs fixes that seemed relevant. All the while there were changes to UUID formats (but not consistently), additional tables being created in the databases and changes to the voicechat server which caused plenty of consternation. I noticed last night that there are patches adding an extra column to a table that was added after the 0.31 release.
This week, just prior to the Innovation Showcase I built a server from SVN version 445 and discovered that voice didn’t work properly if you had more than 1 region in a simulator. I trimmed my sails and deleted two of the regions since I only needed one to demonstrate the support center concept.
Given all that, would I recommend that you wait for realXtend to stabilize before you start experimenting with it? Probably not, but I wouldn’t set a “go live” date for a project either. There are just so many unknowns in the world of Opensim & realXtend. For example, the permissions model is not working yet. It’s hard to imagine going live when you can’t set permissions properly and be sure that new users can’t inadvertently damage something. Likewise, it’s hard to do collaborative development when you can’t edit your team member’s script inside an object. Performance at scale and network latency have not been measured with the various database options. Operational frameworks for creating users, changing passwords, managing test and production servers, etc., are practically non-existent. Even the basic procedures for upgrading from one revision to the next have to learned the hard way. (Owners of public grids are creating a variety of tools to manage their grids but a lot of this innovation is not flowing back to the core distributions.)
Bottom line: go into this with your eyes open. You are a pioneer. Your contributions through testing, bug reports, questions on forums, blogs and tweets can all help to refine the platforms and make them more stable. If you’re working on this stuff in your spare time without a development team supporting you, don’t set any expectations about when your project might be ready.