No rhyme or reason, just wanted a change says Linus Torvalds when commiting the next version of the Linux kernel from 2.6.40 directly to Linux 3.0. This is great news in a way as the 2.6 seemed an infinite series, which never was going to end.
This thread of renumbering was discussed at the Kernel Summit last year and was to be revisited this year too. However, Linus seems to have pulled rank and gone ahead with the renaming, promising the number-change will bode good for all.
It is just a simplification of the numbering process and is definitely not to indicate a platform altering changes to the kernel. In comparison, the previous version, 2.6.33 seems a more mature and better-improved version. Neither is it a feature-jump as Gnome did, from its Gnome 2.6.30 to Gnome 3.0.
Beginning its journey (almost at the beginning of Linux) in 2003, the Kernel 2.6.0 is definitely not, what the 2.6 kernel level is today.
In his Linux Kernel Mailing List, Linus comments that Linux Kernel 3.0 will not have any linux-dom shattering features but a regular update with the usual two-third driver tweaks and some additional plodding. Some of the VFS too has been cleaned up with fixing of several VMs. Some long awaited initial ARM consolidation. The aim of numbering it 3.0.0-rcl is to for cresting the 6-7 weeks for submissions, script-cleaning, while the third number is for the –stable team to use to keep tab of the revisions they will offer. The final release will be Linux-kernel 3.0.
The renumbering is neither a time-based release, as some perceive it to be to commemorate the ’20 years’, it remains simplifying the numbering for all people involved. Therefore, no new API changes nor ABI changes just a simple short, stable release thanks to the shorter merge window. News, is that Linux kernel 3.0 will not have the standard “.0” hiccups as during other projects.
While the common conclusion after every one has accepted the renumbering is that this will not bother the regular end user. This is largely because the distribution of packaging kernels is usually without a reference to the Linux Version number. It is the kernel developers who will have to think on their feet to get over the double-stepping of the kernel.
Another view point is that the new numbering could attract new users as well as it more likely that users would simply assume a higher numbered kernel is the latest version rather than being a driver changer without even API changes ! However, it has to be admitted that the 2.6.0 is no match to what the 2.6.33 is capable of.
Bringing in the new, for novelty, is highly recommended by the Linux world and as always Linus is probably the best judge on why skipping numbers is the best step forward for Linux Kernel 3.0!