Lack of testing 'threatening stability of Linux'

A lack of commitment to testing by the Linux community may ultimately threaten the stability of the operating system, Linux kernel co-maintainer Andrew Morton has warned.

Speaking at Linux.conf.au 2005 in Canberra, Morton said more attention needed to be paid to testing to avoid bugs proliferating in the kernel, which forms the basis for commercial Linux distributions as well as being used directly by many open source enthusiasts.

A key challenge is the lack of recognition for people who spend hours testing new kernel releases.

"They get no thanks or credit or money... or anything," he said.

Morton also criticised the Bugzilla tool used for tracking problems, saying that it encouraged one-to-one communication, a process which didn't help educate the wider community about potential problems.

"Bugzilla is fine for tracking bugs, but as it's currently set up, it's not very good for resolving bugs."

"This process, where individuals communicate via a Web site, is very bad for the kernel overall."

The switch from alternating between even-numbered 'stable' releases and odd-numbered 'development' releases with version 2.6 of the kernel has been controversial, but Morton said the process had generally worked well.

"We've made big changes to the standard kernel, and haven't broken it any more than it was before."

"One thing I hope to do over the next year is to settle the development process down a bit," he said. Morton commented on the ongoing war of words between Linus Torvalds and Andrew Tridgell over the recent loss of a licence to use BitKeeper to maintain the Linux kernel source tree, saying he had always had reservations over the use of a proprietary tool.

"I was never very happy over the choice of BitKeeper," he said, saying that the technology had been chosen without sufficient consideration of alternatives or consequences.

"If you pick a good technology and the developers are insane, it's all going to come to tears."

A better approach when BitKeeper was first adopted would have been to back a group of open source developers to create an SCM tool and then adopting that for kernel work, he argued.

"[If we'd done that], the free software world would have had significantly better tools than we do have."

Morton stressed he was trying to avoid direct involvement in the dispute.

"I'm basically keeping my head down until the dust settles and the shrapnel stops flying," he said.

However, he added that the role of SCM had been overestimated in many cases.

"It's only Linus who is directly affected by the BitKeeper business," he said, noting that many projects would need a different approach.

"If you think you need a source code management tool for your free software project, you might be wrong."

Morton also emphasised that he didn't agree with Torvalds' longstanding philosophy that rejecting patches from the kernel was just as important as accepting them.

"I diametrically disagree with him on that stuff," he said.

"If we just drop the patch on the floor . . . it's the kernel that ends up missing out. It's the maintainers' function to get patches into the kernel, rather than taking pride in rejecting them."

Angus Kidman reported from Sydney for ZDNet Australia. For more ZDNet Australia stories, click here.