Wednesday, April 09, 2008

Linux Foundation Collaboration Summit, Austin 2008, Day #1

Have you ever had those weeks where your calendar is continuously double booked, or triple booked? Well, this week has been double booked solidly and quadruply booked sometimes, with the challenge of having some of those meetings being in different cities. This time the cities for my meetings included Beaverton (close to my nice home office in Portland, Oregon), Orlando, and Austin. And, all of them are activities that I normally have a strong interest in. One was IBM's Technical Leadership Event in Orlando, which pulls in some of IBM's top talent and provides best of breed internal training. That was a must-attend event, and I was able to catch the last bit of the next event which was also a must-attend event.

The Linux Foundation's Collaboration Summit, began on Tuesday in Austin. While I wasn't there, at the beginning, many of my counterparts were able to both represent and take some notes so I can get back up to speed. I'll share some of those here since there is some good stuff in here.

First, the program started with the "State of the Kernel" panel. One of the topics focused on kernel quality, where the kernel community seems to see all the problems - there is a lot less visibility to the good things about any given kernel release. However, even without balanced input, the consensus seems to be that the kernel quality has improved over the past year. My own experience seems to reflect that as well in that most kernel updates lately seem to demonstrate that the tricky things like power management, suspend, hibernate, laptop drivers, etc. Normally, I'd check some of the automated kernel test results to help validate my thinking, but it looks like the conversion to autotest has left the answer a little less clear then they used to be but this view comes closer to providing a snapshot of kernel stability. But at a broad subjective level, that consensus appears to reflect the broader consensus that I typically see.

Device drivers were again a topic, although in the past year, a phenomonal step forward in terms of the transitioning from perception of driver problems by the user community and perception of phenomenal driver coverage by the development community (both perceptions were right, of course!) with the creation of the Linux Driver Project and their list of drivers needing love the problem has transitioned from a near pointless debate to a focused activity where users and developers come together to solve the problem. The downside is that there are still a lot of drivers needed but the good news is that people are steadily working their way through the list.

There was a question regarding the Linux Test Project and its interaction with the Linux Kernel Community - the key problem is that LTP sometimes is out of date, covers surface tests, or generates false failures. I know the LTP team is working on that, and the kernel team is interested on working through that. There was a proposal at the kernel summit to include LTP in the kernel tree to help fix and maintain it more aggressively. However, Linus was not too keen on that, so that has been blocked. I don't have a current update on where that sits but would love to know the latest thinking.

And, last for that session, RealTime Linux was cited as a great example of the community coming together to find a consensus solution amongst competing alternative activities and potential solutions. RT is now used in production.

That's probably enough for one entry... Look for more as I find time.

Labels:

0 Comments:

Post a Comment

Links to this post:

Create a Link

<< Home