[zanny]

I wouldn't be surprised if Linus pushes for 5.0 after 4.9 this time. And then maybe after that you realize the major version number is meaningless when you will never break backwards compatibility (like with web browsers) and drop it entirely.

So maybe Linux 30 by 2020?

[nirvdrum]

Web browsers break backwards-compatibility all the time. They use the major version number bump to justify it. I gave up trying to keep a Firefox extension I wrote up-to-date since the API changed every 6 weeks. I think it's settled down a bit since the early days, but the reason they bump the major version number is so they can break things.

[zanny]

I meant more in terms of they don't break the standards they implement, they just add more on, but yeah mechanically its a misnomer.

[nirvdrum]

Gotcha. Outside of extension devs, it probably doesn't matter. But I figured I'd mention it.

[vacri]

He can't drop the decimal easily, as there are apparently too many scripts that expect the kernel version to have a decimal in it. But I think 4.9 -> 5.0 is a decent workaround to the problem.

[zak_mc_kracken]

That's funny, because the reason why Microsoft decided to skip from Windows 8 to Windows 10 is because there's too much software out there that tests that the version of Windows starts with "9" and interprets this as they are currently running Windows 95 or 98.

Some things never change.

[SquareWheel]

Highly improbable to be the case. Microsoft damaged XP compatibility with Vista; that they would care about software built for 95-98 in 2015 seems unlikely by comparison. It's also just a product name. Programmers will check the kernel version, not the "friendly" consumer-facing string.

I realize this information has been circulated around, but unless this has been confirmed by Microsoft please don't claim facts without evidence.

[JelteF]

I hope you do understand that it would not damaging compatibility with 95-98, but with windows 8. The class of programs that this is for runs fine on Windows 8, because they are not executing legacy code for 95-98. But the check for 95-98 would return true for windows 9. So the legacy code will be executed, probably breaking stuff because the API has changed a lot.

And as for some evidence this is the case: https://searchcode.com/?q=if%28version%2Cstartswith%28%22win...

[SquareWheel]

I do understand that. We'd be talking about programs built in the 95-98 era however, and the majority would not be in current use. The biggest exception would be enterprise, though if they're upgrading to a recently released Windows version they'd (hopefully!) be using more modern software too.

At the end of the day, we're still only guessing at Microsoft's rationale. Esoteric compatibility issues seem lower down the list from issues like breaking away from the Windows 8 name (which has accumulated negative association in the public view).

Until Microsoft says for sure, it's just a bit silly to make the claim as fact.

[anon4]

Microsoft would never say for sure.

And you're discounting company-internal products that were built in the 95-98 era, then updated when new business regulations came out, but were never rewritten in a modern language and the company wouldn't spend time and money to port them to Windows 9 when they worked fine on every previous version -- they'll just wait until a newer version of Windows fixes the problem or Microsoft very forcibly drop support for 8 completely.

[zak_mc_kracken]

> Programmers will check the kernel version, not the "friendly" consumer-facing string.

Cute. New to the profession?

Chew on this:

http://www.reddit.com/r/technology/comments/2hwlrk/new_windo...

[SquareWheel]

A deleted comment by somebody claiming to be a Microsoft developer, commenting on an "internal rumours". That's not what I would consider "evidence".

And there's no need to use the condescending "cute". I certainly wasn't trying to start a flame war here. I'm asking you to verify information before repeating it as fact.

[girvo]

> Programmers will check the kernel version, not the "friendly" consumer-facing string.

I wish that were true, and it may be if you're dealing with competent developers, but I've seen code that does exactly that in the wild!

[brudgers]

Here's some evidence to support the inference regarding the out of order naming:

https://searchcode.com/?q=if%28version%2Cstartswith%28%22win...

[amyjess]

In one of those examples:

> if (osName.indexOf("9") != -1) {

Jesus Hermann Christ. It's not just looking for "Windows 9", it's looking for a "9" anywhere in the version string. Yes, it's under an if clause looking for "Windows" also in the version string, so it won't trigger on "Plan 9" or something, but still it'll get triggered by something like Windows 19.

[zak_mc_kracken]

Interesting that you would call my claim "highly improbable" because I didn't cite any sources and in the same breath, make a claim without any sources.

"Unverified"? Sure. "Highly improbable"? You don't know that any better than I do.