Thursday Oct 25, 2007

The impression of the OpenSolaris Developers summit 2007 (2)

Then how about the summit itself? It's well-organized and the facilities are nice. Thanks for the team, especially Jessy.

As the first priority task for the next few months is Indiana, Indiana is the focus for the first day. The welcome session was delivered by Ian Murdock. He gave a general talk on the topics we were going to discuss: the distribution model, the new installer, the new package system, modernization and the distribution constructor. This is the first time I met Ian. When Debian came out as a Linux distribution, the apt-get command changed our life. I can still remember the Debian related discussion between some friends and I back in the university:-)

The round table introduction session was great, with everyone involved, and I love this part. It set the fundamental key that people in this community could have a free platform to communicate, without any title. People from Sun(management team and engineers) showed their participation enthusiasm, interests, attentions and supports to the future development of OpenSolaris community. The community contributors also expressed their ideas, suggestions and concerns. The summit opens a channel for the communication between people from Sun and people in the community. And we were trying to build up some common understanding about this community.

Here I'd like to drag the topic a little bit far away. As an engineer works for Sun and joins an external community(Mozilla community), I prefer influence than control over an open source community. Different parties have their influence can help the community to grow in a benign way. If we want to make the decisions based on democracy, communication becomes the most important thing. I believe that's the purpose of the summit.

Following are the sessions, some with slides, audio recording and meeting minutes:
Stephen Hahn: Image Packaging System slides audio minutes
When I migrated from Linux to Solaris, one thing I wanted to have is just the packaging system. Blastware partly resolves the problem at that time. Just like the apt-get changes our life, this new image packaging system could be the key point to make Solaris much more user/developer friendly.

Dave Miner: Installation and distribution constructor slides audio minutes
It was mentioned that Developers will be the first priority, and enterprises later for Indiana. That's a great news for all the open source communities I think. The LiveCD demo was also good. BTW, the desktop China team also gets involved in the Caiman project, which is already in SXDE III.

David Comay: Modernization audio minutes
This brought some hot discussions in the OpenSolaris community before. Familiarity means a lot of things: to remove the outdated legacy commands, the obsolete directories, to make the system simple to configure, to make the development environment easily available. To provide a familiar system to Linux users/developers could make Solaris much more competitive from my perspective.

Several issues were raised during the meeting:
1. How to prevent fragmentation for OpenSolaris community? Different kinds of fragmentations exist in the history, BSD(FreeBSD/OpenBSD...), Linux(RH/Ubuntu/Novell...), Ubuntu(desktop/education...). The answer to this is that the repository of the image packaging system will be the source for all the distros to derive from. Solaris ON(OS and Network) and a set of core functions will be a must for distros to make sure binary compatibility. Within this limitation, fragmentation could be a benign thing for the community development. But this could still be a problem to pay attention to in future development.

2. Solaris has a strong backward compatibility guaranty. How will Indiana address this issue? The discussion was around whether it's important for attracting customers to have the backward compatibility. It's mentioned that Indiana only keep partly backward compatibility. Actually, the key idea is that different areas have different backward compatibility requirement. For example, I don't think it's so necessary for us to put a strong backward compatibility requirement to the desktop applications. But for Solaris ON, this is still important. This kind of loose statement could make the Solaris a more fashion look, especially for desktop environment.

3. Time to market. This is more about the release schedule I think. Currently, Solaris has a long development cycle. Solaris 10 is quite old(but stable), especially for developers. Indiana could provide a release vehicle to host all the latest features/applications. By a 6 months release schedule, it can shorten the time to market and also help collect some customers' feedback in a timely way. Then our development could benefit from this as well. A win/win situation.

The aboves are almost all the contents for the first day. It's quite helpful for me to know about these. Solaris is competitive to compare with Linux/Mac/Windows from my point of view, as a developer. It has so many features that the other systems don't have. Just like DTrace can help Mozilla in my previous posts, the other open source communities can also benefit from it. All the changes Indiana will bring to us are the key points for Solaris adoption. And I truly hope that it can be available ASAP and I can then recommend it to some other guys around, realistically or virtualizedly.

The party at night was also wonderful. Emily recommended me to have some Guinness. My impression is it's so different:-) Jack is funny. It's cold outside, but warm when close to the heater. The toast is short. I believe that we could have more toast when Indiana is available.

BTW, why is the code name Indiana? Is it because that Ian lives in Indiana?


« August 2016