Website Transition Update

As we get closer to our transition to the new infrastructure, we will be posting documents to the Website community to explain the migration plans. We also want all this information to start growing in one place so it's a convenient reference point for the community. And we'll be posting announcements to opensolaris-announce and website-discuss to engage the community in the implementation process.

There will be content migration tests as well as a period of time when we actually transition to the new infrastructure. So, thousands of people around the world will have to be notified and engaged as we move. Should be an interesting exercise, eh? We'll do the heavy lifting by migrating the content and user data, but there will be some clean up issues along the way and some significant changes that everyone will have to get used to. There will be many cool new features we'll get, of course, and there will be some old features we'll not bring over initially or at all. It all depends on use patterns and the natural opportunities and restrictions of the new system (which is inherent in any system). Most of the user data migration challenges have been worked out, and we are finalizing the content migration and translation tools now. We have the ability to move all of the site's content now and translate it from TML Markup and/or HTML into XWiki Markup. This is cool because it will enable people to use the XWiki WYSIWYG editor instead of messing around with wiki markup tags or HTML. But we still have some formatting issues to work out and a bunch of skinning and information architecture things to do as well. More on all this later.

Development of the new infrastructure has been going on for some time now, and we`ve been having conversations about it at various points on website-discuss as well. Currently, most of the core engineering work is complete, so we are starting to implement this thing in a series of steps. The immediately goal is to figure out the absolute minimum set of implementation requirements to enable us to transition to the new infrastructure as quickly as possible. This is a critical point. We grew well beyond the capabilities of current website long ago, and we need to move to the new system to continuing growing (and also to stop crashing as we have been doing lately). But once that initial transition is complete and everyone in the community is working on XWiki, we will not be done. Far from it. That`s just the first step (albeit a big one). There are other applications supporting open development that will interact with the new authentication database under XWiki and they need work and migration, we have to solve the list/forum gateway problems, the site will need to be localized into many languages, there will be user education challenges to overcome, the new site will have to be evolved as any new application would, and the list grows. But you have to start somewhere, right?

Keep in mind we are changing the engines while plane is jetting thorough the sky, so this could be a tad complex from a technical, organizational, and social point of view. Please have patience. :) We will make mistakes. Things will break. But we'll fix them and move on as fast as we can. And in the end, we'll have a much more secure and flexible website. But remember, we are not just building a website here. If that were the case, we'd be done long go. Instead, we are building an entirely new platform to support open development for the OpenSolaris engineering operation -- and that's a distinction with a very big difference. Those open development bits will emerge over time as we move infrastructure outside, but all of this is designed to support the growth of a global engineering community. We are trying to think big and long term but also build things in small pieces over time.

Anyway, below are some documents outlining some of the aspects to this transition. Some of these documents have been around for a while, but the user data migration strategy is new as of this week. Give it a read to understand how we are pulling your user data from multiple locations to centralize it in the new system so it is expressed the same way across multiple applications.

Working on this website transition project is where I spend virtually all of my time now. Alan Burlison is the tech lead. I'm serving as the program manager. And the others doing the work include: Chris Phelan, Martin Walsh, Bill Rushmore, Derek Cicero. We are part of Bonnie Corwin's Developer Collaboration Engineering Team, so she'll be involved all along as well.

More info as we go. We're getting close.


What stage is the Jive replacement at? I think the idea of breaking the link between the web forums and the mailing lists is sound, however there still needs to be prominent links to the mailing list archives on the website. Also, we still need web forums - in my view mailing lists are not a replacement as they tend to be less visible and I consider them to be less "newb friendly" than a web front-end.

I think the decision to go for a web forum or a mailing list needs to be done on a list by list basis - there should not be a wholesale switchoff of one side or the other. Some forums are better online, some better on email. I expect the more developer orientated lists/forums to go down the mailing list route and the lower traffic developer and user lists to remain on a web forum.

Just me tuppence worth!



Posted by andrewk8 on March 18, 2009 at 11:53 AM JST #

It`s not on the schedule yet, but we know we have to address it. We`ll start a requirements discussion at some point very soon (just as we did with the SCM and Wiki issues). I agree we can`t continue to scale with having web forum gateways to \*all\* the lists. We have over 350 lists about about 150 forums, and we are no longer even creating forums at all until we can fix the situation. Part of the problem is that Jive is integrated so tightly into the webapp, so we have very little flexibility at the moment. Also, it can`t handle the number of gateways we have. So, after we move to the new infrastructure, we can pull the lists and forums apart and dump the gateways. Or we can move to an entirely new forum system, which if we do that we are likely to just skip the gateway issue (my opinion only at this point). I would like to go with an open source solution if we can, and use forums for more user-oriented functions and lists for developer-oriented functions.

Posted by Jim Grisanzio on March 18, 2009 at 01:50 PM JST #

Post a Comment:
Comments are closed for this entry.


« June 2016

No bookmarks in folder