The 2003 Kernel Developers Summit [LWN.net]

[Reporting by LWN.net Executive Editor Jonathan Corbet]

The 2003 Kernel Developers summit was held July 21 and 22 in Ottawa, immediately before the Ottawa Linux Symposium. Nearly 70 developers attended the invitation-only event and discussed numerous topics relevant to the ongoing development of the Linux kernel. LWN editor Jonathan Corbet was there, and has documented most of the sessions below.

Monday sessions

The requirements panel; a six-member panel representing large Linux vendors and users discussed their wishes for future Linux kernel developments.
High availability; a session led by Lars Marowsky-Brée on what Linux needs to better support high-availability applications.
Workload management, led by Ken Rosendal.
NUMA management, led by Andi Kleen.
Error and event logging, a session led by Suparna Bhattacharya.
I/O clustering; Gerrit Huizenga ran a session on improving file write performance.
ACPI and power management, a discussion of power issues, hardware management, and sysfs led by Andrew Grover and Pat Mochel.
Greg Kroah-Hartman discussed his udev utility as a way of replacing devfs. LWN covered udev back in April, and it has changed little since. About the only news was the announcement that a persistent device naming and permissions scheme is in the works and will be added soon.
Five-minute brainstorming; a free-form session where kernel developers could quickly present ideas or concerns. This session ended with Linus talking a bit about the 2.6.0 release and the eventual 2.7 fork.

Tuesday sessions

Tuesday's sessions included:

The Processor Architects' Panel brought together processor designers from three manufacturers to talk about where their product lines are going and implications for the Linux kernel.
Transparent superpages and page clustering, discussions led by Dave Mosberger and William Irwin.
Reverse-mapping VM performance, a session led by Rik van Riel.
Martin Bligh ran a session on various VM topics.
Asynchronous I/O, run by Suparna Bhattacharya.
James Bottomley celebrated his birthday with this session on the SCSI subsystem.
Multipath I/O; this session run by Mike Anderson and Lars Marowsky-Brée looked mostly at problems with current Linux multipath implementations and the desired features for a new implementation.
The closing session of the summit was a discussion of what went well (or not) with the 2.5 development series, along with a look forward to how 2.6 will work.

The traditional kernel summit group photo is also available.

(Log in to post comments)

Let's not forget

Posted Jul 22, 2003 14:38 UTC (Tue) by dwalters (subscriber, #4207) [Link]

LOL!

In all seriousness, though, why isn't SCO on the agenda at this summit? I'm sure SCO's actions weighs very heavily on the minds of most of the kernel developers who will be attending the summit, and if the kernel developers put their collective brains (and copyrights) behind it, I'm sure they could come up with some effective counter-attack.

Let's not forget

Posted Jul 22, 2003 17:22 UTC (Tue) by JoeBuck (subscriber, #2330) [Link]

This is the technical summit, not the legal/business issues summit. Still, I'm sure that there will be lots of quiet conversations about what to do. It might be best if the kernel copyright holders empower Linux International (or some other body) to deal with the legal issues, just as Linus has done for trademark enforcement.

At some point, some body with official standing is going to need to contest the issue in the press, and possibly issue official announcements, for example, that SCO is in breach of the GPL and the Linux copyright holders are invoking clause 4, and threaten similar action against anyone who publicly pressures anyone to obtain an SCO license in order to "legally" run Linux (as such pressures are an attempt to sublicense a GPLed work). The kernel developers should be insulated from this ugliness, but we can't let SCO go around extorting licenses from people without objecting.