Linux 3.12 Kernel Released; Linux 4.0 Planning Talked Up

[BSDude]

Actually back in 2011 when Linus announced his desire to end with the 2.6 series and move to 3.0 once the kernel hits its 20th anniversary it seemed to me a good cut-off mark. Why don't they just move to 4.0 once the kernel turns 30? Like 4 would stand for the 4th decade of the linux kernel. Then the point release would stand for the year it was released from 0 to 9 and then the third point release would stand for all the sequential kernel releases in that year.

So it would look like this

a.b.c

where 'a' would stand for a release in the 2020s, the second 'b' would stand for the fourth year of the 20s (ie. 2024) and the 'c' would signify the second stable release of the kernel in 2024. I would assume that there would be no more than 5 releases in a particular year. If a new kernel is started at the end of the previous year and is released in the next then 'b' is bumped to the next number and 'c' goes back to 1.

This versioning scheme definitely won't reach crazy numbers. We definitely will be long gone by the time the kernel will be in the twenties The negative aspect is the lack of simplicity at a glance, it has more dots. Also there is this discrepancy between the actual anniversary being in 1991 which puts the previous year in a bit of dilemma but I think that can be ignored and just consider the entire decade from 0 to 9 as to how many years the kernel exists.

[ObiWan]

Well 2016 is the 25th birthday, so he could use that as excuse for a version up.

[GreatEmerald]

A bugfix kernel would at least set 4.0 apart, and would be a pretty good very-long-term-support release. But then the next kernel version would explode with features...

[d2kx]

Final Linux 3.12 kernel runs great here with AMD Catalyst 13.11 Beta 6 (Ubuntu Mainline PPA kernel). Performance (general, haven't tried 3D) is awesome with my AMD E-350 APU!

[drev]

Linus doesn't want to have a Linux 3.(some-large-number)

what's wrong with that ?

[s_j_newbury]

Actually back in 2011 when Linus announced his desire to end with the 2.6 series and move to 3.0 once the kernel hits its 20th anniversary it seemed to me a good cut-off mark. Why don't they just move to 4.0 once the kernel turns 30? Like 4 would stand for the 4th decade of the linux kernel. Then the point release would stand for the year it was released from 0 to 9 and then the third point release would stand for all the sequential kernel releases in that year.

So it would look like this

a.b.c

where 'a' would stand for a release in the 2020s, the second 'b' would stand for the fourth year of the 20s (ie. 2024) and the 'c' would signify the second stable release of the kernel in 2024. I would assume that there would be no more than 5 releases in a particular year. If a new kernel is started at the end of the previous year and is released in the next then 'b' is bumped to the next number and 'c' goes back to 1.

This versioning scheme definitely won't reach crazy numbers. We definitely will be long gone by the time the kernel will be in the twenties The negative aspect is the lack of simplicity at a glance, it has more dots. Also there is this discrepancy between the actual anniversary being in 1991 which puts the previous year in a bit of dilemma but I think that can be ignored and just consider the entire decade from 0 to 9 as to how many years the kernel exists.

Of course you realise, decades, centuries etc, technically end/begin on year 1.

We're currently in the second decade of the 21st century, and will be until Jan 2021, 2020 being the last year of the second decade. This irritated the hell out of me with the turn of century...

/pedantic_mode

[s_j_newbury]

http://www.loc.gov/rr/scitech/battle.html

[leif81]

For a guy who runs such a tight ship it's pretty surprising how useless the Linux versing numbers are.

I personally really like semantic versioning http://semver.org

[arti]

Linux 4.0

So, some time in the future we will have Linux NT 4.0 ?

[ciplogic]

if they just plan 19 releases of a kernel version and the 20th is a new one, maybe they should make it more predictable.

for example from v4 to v5, it would be more predictable to go by multiples of 5:

4.00 , 4.05, 4.10, 4.15, 4.20 [...] 4.85, 4.90, 4.95, 5.00


going from 2.60 to 3.00 and from 3.19 to 4.00, without it being an "awesome" big release and just a normal one is kinda weird IMO.

But who said FOSS has ever been "predictable" ...

Who said it should be predictable? Which software package is 100%?

GCC 4 was a major rewrite of how optimizer works and for at least 1 or maybe two revisions was catching up to version 3. KDE4 was not predictable and Gnome 3.0 certainly wasn't. So why expecting something predictable from OSS?

But who expects predictable releases, is mostly because wants the same stuff but more polished. Linux kernel is a mature stuff but it also active as a project, so I do see the point that from time to time to update the version to think as "feature levels".

But compared with real software like Windows, which in one year they make the "desktop is a tile", and anyone should write WinRT, Linux is more predictable. GTK+ 2 to 3 was a minor change by all standards, KDE 3 to 4 too.