The management does not end there: the folk at Ubuntu are also very good at keeping the distribution's name in the headlines.
A recent case that illustrates this is the blog post about the btrfs filesystem by Ubuntu developer Scott James Remnant. The post was headlined "btrfs by default in Maverick?", the reference being to the next release of Ubuntu, Maverick Meerkat.
This headline has been repeated in various publications around the net and I've been waiting to see if anyone will have a second look at Remnant's post and dissect it in detail.
Nobody has bothered to do this, even though the post is all of 10 days old. Nobody has bothered to even recall that it was only very recently that ext4 became the default filesystem in Ubuntu. Why would one want to make what is a major change so soon?
If one reads carefully through what Remnant has written, there are lots of riders attached. Some of the conditions he lists are the removal of the "experimental" label on btrfs (which is expected in the 2.6.35 kernel release), and support for using btrfs with GRUB2.
But even if these conditions are satisfied, all that Remnant states is that the Ubuntu team may make btrfs the default filesystem for alpha releases! (emphasis mine). That's all. However, he had no hesitation, in headlining such a speculative post "btrfs by default in Maverick?"
The post was well-timed - it came soon after the blizzard of publicity about the last Ubuntu release was dying down. Excellent timing to keep the distribution in the headlines by flying a kite that has very shaky wings.
As one senior FOSS commentator pointed out to me: "He said absolutely nothing about making btrfs the default in the beta and final releases. Just in the alpha releases 'to gain more testing'. The way this has been portrayed in the media and other blog posts has consistently gotten this wrong. My, perhaps cynical, suspicion is they want more ad revenue by using the more attention-grabbing (but wrong) wording."
Not that btrfs is a filesystem to be frowned upon. Chris Samuel, who works with the Victorian Life Sciences Computation Initiative at the University of Melbourne, has been using the filesystem on laptops for the /home partition for about 18 months without any issues, and says: "Currently, btrfs works well, but does have some rough edges and (I believe) there are still some ENOSPC edge cases where running the filesystem out of room to manoeuvre can lead to a kernel oops.
"Also, its inbuilt RAID is currently limited to RAID-1, though there are out-of-tree patches in development for more RAID-5 and 6."
Samuel, who has considerable experience with filesystems and was formerly the systems manager for the Victorian Partnership for Advanced Computing, points out that in the case of his laptops which run btrfs for /home, he has plenty of spare room.
He says the advantages of running it are "the possibility of rollback for package installations (though that would mean a lot of fundamental work in apt/deb), a filesystem that should always be consistent, checksumming of data on disk so you can spot corruption (and possibly recover automatically if you've got the mirroring), etc.."
On the downside, Samuel points to "the new, potential ENOSPC crashes (if that doesn't get nailed down fully beforehand), only inbuilt RAID-1 support (of course you can use it with software RAID), other edge cases that haven't been experienced yet due to not being in widespread use."
When I interviewed Linux filesystem guru Ted T'so earlier this year, he was at pains to point out that it would take something like 100 man-years for a filesystem to be built from scratch to production-readiness. As an example, he said, Sun had something like 12 people working for five years on their ZFS before they released it in 2005.
Those who blithely propagate comments about changing default filesystems would do well to adopt some of his caution.