Linus Torvalds Still Deciding Linux 3.20 vs. Linux 4.0

[YoungManKlaus]

just stick a date on it

make it 2015-02 and be done with it. Maintenance releases are then like 2015-02.01 and so on.

[dungeon]

make it 2015-02 and be done with it. Maintenance releases are then like 2015-02.01 and so on.

That is boring, read this:

As of writing, 57% are voting for "v4.0, 'cause I get confused easily" while 43% voted for "I like big versions, and I cannot lie."

The best and popular products on planet Earth has always exact fifty-fifty of those who agree and those who does not

[gnufreex]

I voted 4.0

[dungeon]

Oh my, linux will get 4.0 before Mesa

[carewolf]

This is why he should have gone with 3.0.x, so this would only be moving to 3.1.x.

Bumping major version number for no reason is just obscene. Last time he atleast had the excuse of adopting a new version system. He just adopted a bad one, which now leads to more pointless breakage.

[duby229]

This is why he should have gone with 3.0.x, so this would only be moving to 3.1.x.

Bumping major version number for no reason is just obscene. Last time he atleast had the excuse of adopting a new version system. He just adopted a bad one, which now leads to more pointless breakage.

Well, that's partly true. But at this pace it's going to take decades before we hit something obscene like v12.0. It'll take a long time before major version numbers get too high.

[bregma]

Linux is so behind

Microsoft Windows is closing in on version 10. Mac OS has been at version 10 (with hip Roman numerals no less) for years. Linux is mooting releasing version 4. That's 6 whole versions behind, who would want to run that old crap?

Linus should pull a Volkerding and jump straight to Linux kernel version 11.0.

[Ericg]

This is why he should have gone with 3.0.x, so this would only be moving to 3.1.x.

Bumping major version number for no reason is just obscene. Last time he atleast had the excuse of adopting a new version system. He just adopted a bad one, which now leads to more pointless breakage.

Version numbers as a whole are completely arbitrary anyway, they matter yes (in relation to eachother) but they as a whole scheme are arbitrary. There's no "pointless breakage" either, apps that only check for "2.6.x" or "2.4.x" broke anyway during the shift to 3.x. Any app written to take account for 3.0 should've been written to take into account for the fact that the version number is subject to change as shown by the move from 2.6.x to 3.x. If anything breaks at this point it is completely the developer's fault as they were warned.

I really don't get all the hate that surrounds version numbers, whether it be the kernel or anything else. We've hit the point technologically where there aren't revolutionary releases and everything is more evolutionary-- release early and release often. Plus the benefit of Linus' "Dont break shit" policy in the kernel means that there would likely never be a "We broke everything. Major version bump." release anyway. The last time that happened was like 2004 during the switch from 2.4 to 2.6-- 11yrs ago. (God 2004 was 11 years ago...)

[schmidtbag]

I dont understand why they use this versioning scheme if it does not fit the purpose. Why major version gets bumped if there are no big/breaking changes is beyond me. Confusing numbers is lame excuse... Since kernel development is incremental why dont they just adopt same version numbering like ff/chrome/systemd...

I agree. TBH, I think the 3.0 kernel was a bit late. That should have occurred when HAL was officially depreciated, because (IIRC) that actually caused some regressions in some cases due to driver compatibility. That's a solid reason to do a major version change. I'm not a fan of bumping up a major version "because the minor version is getting a little high". In a hypothetical situation, that's like if Volkswagen renamed next year's "Beetle" to "Insect" without any major changes to the design, when there have been models where they changed the fuel type or engine size and stuck with the old name; those things are significant enough to differentiate.

In other words - if the Linux 4.0 kernel has no major change, what will they do if a major change happens? Technically, the linux kernel is still largely compatible with the 2.6 kernel. You shouldn't just arbitrarily change the version just because "it feels right".

Linux needs a new versioning scheme, because at this point it is just nonsense and based on a whim. Every once in a while, Torvalds' spontaneous attitude takes away his genious.

[Ericg]

In other words - if the Linux 4.0 kernel has no major change, what will they do if a major change happens? Technically, the linux kernel is still largely compatible with the 2.6 kernel. You shouldn't just arbitrarily change the version just because "it feels right".

You mean the inclusion of Kdbus? Swapping out iptables for nftables? Splitting GPU drivers into render-nodes? "Major" changes don't just get tossed in and happen all at once, they happen incrementally and gradually sneaking in bit by bit under the radar.