Slashdot Mirror
Linus Torvalds Says Linux 4.0 Could Be Out In Three Years
darthcamaro writes "The wait between Linux 2.x and 3.x was a long one, but the wait to Linux 4? Well, that will only be a matter of three years, according to Linus Torvalds. '"It's just mentally much easier for people to remember the small number," Torvalds said during the LinuxCon conference in San Diego [Wednesday]. "We'll do 4.0 in three years maybe when the sub numbers have grown in the 20's and our feeble brains can't handle it."'"
Firefox will be up to 1,376,265.1 by then.
Also, not covered in the summary, just like the GCC 4.X series, it will finally compile in C++.
</joke target="for the impaired">
SJW n. One who posts facts.
These days it's all about dumb terminals and VAXclusters.
...that age is just a number?
Alternatively: "Life begins at 4.0".
Judging from 3.0 which didn't have any breakthrough features included, this is just silly numbers talk.
They're merely version numbers, after all.
It's just mentally much easier for people to remember the small number,"
How about 0.00000000000000000000000000000000000000000000001?
.
Captcha: impudent - is Slashdot trying to tell me something?
Most Linus users don't know their kernel version anyways. They just know their distro, and maybe distro version, and never care to look at what is under the hood.
Usability be damned, I would prefer they encode the version number in I's,N's, and U's. Running kernel version Liiinnnnnnuuux.
while(1) attack(People.Sandy);
We'll do 4.0 in three years maybe when the sub numbers have grown in the 20's and our feeble brains can't handle it.
If your numbers are going to be arbitrary, why not roll them over at 3.9?
Focusing on parallel execution and efficient scheduling of large number of processors.
Scheduling is now a rather complex item requiring more than just memory+ready to run.
Memory (where is the memory in a distributed system).
ready to run (where is the available processor)
scheduling additional constraints such as communication delays between memory and processor, between processor and peripheral, between peripheral and memory (DMA).
How to compute appropriate weighting efficiently, and fast.
Detecting complex distributed deadlocks, and determining recovery strategy with a minimum of computation time lost.
It gets much more complicated with such poorly designed architectures such as the X86.
What would be a better design for distributed systems? What kind of network should be used? What kind of granularity in scheduling is needed? What should be doing the scheduling ? Hardware, as in torus designs? or bus switch now that multimode fiber makes serial computing fast again?
What kinds of OS for a serial processor (or a optical processor where inputs strictly come from an input stream and continue to a separate output stream) should be used?
Lots of questions.
FYI, Linux uses git, which uses a SHA-1 as a commit number, not incremental revision numbers.
Also, it's just as easy to remember that 3.26 -> 4.0.
So they went to 2.6 for the previous major version and now they're going to 3.30? How is that not a longer wait?
When will this quick versioning madness end?!!?
internet links are to slow for that and caps will kill that idea.
The kernel has been in a very good shape for a long time already. It's already a "It Just Works(TM)" thing. The aspects that I am interested seeing advancing are in the userspace: desktop environment and games.
Linux 5.0 could be out tomorrow. Linus just has to say "this is 5.0". What does that tell us ? Absolutely nothing. Sorry Slashdot, I expected a bit better from you.
Does this make any sense really? We thought there will be a move from 2.6 to 2.8, all-of-a-sudden we had version 3.0 (Where are my .4 worth of upgrades BTW?)
How much time did it take to move from 2.6 to 3.0? Considering the current, latest kernel is 3.5, it could be decided tomorrow that the next update will warrant a version 4.0. What does this version business equate to? how can you measure how much better it is based on this "version"?
Would it not make more sense to date stamp the release? At least that way you'd know that X development time was put in between 3.5.1 & 3.5.2. I think we need a better system than "version".
A 'singular oddity' is an event that cannot be explained and only happens when you are alone.
Linus gives a quick rundown of kernel version numbering.
The upside is even.x.x means Linus is no longer crazy and he might revert to the slower version rollout.
But something tells me getting from 4.20 to 4.21 will take a really long time, man.
I am officially gone from
I'm sure "new" features will be added, but they won't be tied to any particular major number upgrade. This has been the way OpenBSD has been numbering its releases. OpenBSD 4.9 is simply the version that came before OBSD 5.0, which is the version that came before the current 5.1 release.
Maybe Linus wants to catch up to Theo? Linux kernel releases occur twice as fast as OpenBSD releases, so who knows. I kind of prefer the Ubuntu numeric versioning scheme that lets you know at a glance how old a release is. The animal names though are just plain silly.
For comparison look at the way Microsoft numbers its OS products, and you have to wonder what series they are using: 1, 2, 3, 95, 98, 2000, 7, followed by 8. Maybe they'll call release 9, Windows 2020?
Apple has been stuck at OSX for over at decade now. Two more decades and they'll be triple X.
what happened to the days when 2.0 meant that it was either a complete rewrite or it had significant changes that made it incompatible with 1.x and had undergone extensive testing to qualify it as a "stable" release?
"Insightful?" Not "Funny?"
Palm trees and 8
...or just another kernel version number for servers?
And where will Testing be?
Best Slashdot Co
Warning: This extension is only supported by the latest versions of Firefox-616.
I hope they'd drop the minor version by then... It's getting ridiculous...
Maybe with a MyCleanPC pitch thrown in.
/* No Comment */
The move from 2.6.39 to 3.0 was not a 2.6 to 3.0 move. In effect the idea was that the leading "2" is basically meaningless in the current model, so they chose to reduce it to two numbers instead of three. Once that decision was made, they chose a simple increment from 2 to 3 as the leading digit.
In other words, the number is arbitrary. But there is nothing to be gained by changing something which is arbitrary, by the very definition of arbitrary. Therefore, they could have simply kept with the old versioning plan, which would be equally as arbitrary as the new plan, and saved themselves the effort.
If anyone would have recognized this, it would certainly be a programmer. (D'oh!)
well, what linus seems to have said this time(didn't read the source, of course) is that he'll up the major version once the minor version gets big enough that it's a burden to remember, so it's not totally arbitrary. it's just a number that needs to grow to separate the new version from the old when speaking about it.
world was created 5 seconds before this post as it is.