Hacker News new | ask | show | jobs
by TimTheTinker 2152 days ago
Win98 did though.
1 comments

That's correct. 98 was the first release to have it, chronologically speaking. 2000 was the first NT based Windows to have it.

So 98/NT didn't make sense to me. NT4, the last release to use the name NT, had a very Win95-like appearance. I also remember NT 3.51 (and I guess there were other 3.x) that had a very Win3.1-like UI despite being 32 bit.

Yeah you got me. Fair enough.

Windows 2000 was amazing and just my favorite OS. Saying that as a linux user.

Win2000 was the most perfect and least buggy Windows version of all time.
It didn't seem buggy at the time but you wouldn't want it on the public internet with all its security bugs.

It would be nice if Microsoft just backported security improvements and other kernel type stuff without messing with UI.

Of course but the problem with updates is that you can never only get security updates... You have to get "feature enhencements" too... It would be nice to have 2 branches.
I don't mean just security updates though. I mean systemic security improvements coming from redesigns of components. Some of the changes are feature enhancements in the kernel too, some affecting security.

For all the criticisms of UAC or the Win8+ app store I would sooner trust a sandbox on recent windows than I would on Win2k. Even the XP SP2 era had huge security re-vamps that Win2k never got.

And there are performance things. I happen to know that Win32 message pumps have a better locking scheme on Win7 than they did in 2k.

The issue is in order to get enhancements in the lower layers you get forced into other bloat higher up.

Disclaimer, I spent a few years as a dev on the Windows team.

1.x=1.x

2.x=2.x

3.x,NT3.x=3.x

95,NT4=4.0

98=4.1

ME=4.9

2000=5.0

XP=5.1

2003=5.2

Vista,2008=6.0

7,2008=6.1

8,2012=6.2

8.1=6.3

10,2016,2019=10

Makes perfect sense, right? Honestly I can deal with most of it, but couldn't they have realigned at 7?

https://www.gaijin.at/en/infos/windows-version-numbers

Yeah, in early prerelease they had 7 report as 7.0.

But the story goes there was buggy code out there that said:

    if (major == 6)
    {
       // Do the thing that works on Vista
    }
    else
    {
       // Do the thing that only worked on XP and earlier
    }
And they wanted Win7 to hit the Vista-ready code. So they kept major at 6 for a long time. I guess they stopped caring for Win10.
Even more to the point, the kernel major version number represented an implicit ABI compatibility for drivers.

Windows 4.9 could load drivers for 4.0. 5.2 could load drivers for 5.0. 6.3 could load drivers for 6.0...

Windows 10... can still load 6.x drivers, but it bumped the internal version number to 10.0. It is worked around because without an explicit "I'm Windows 10-compatible!" manifest, the OS just lies and reports an internal version of 6.3, along with the ProductName string remaining "Windows 8.1"

Makes sense, but they also had pretty aggressive detection of when compatibility mode would be helpful, so solving for the bugs with that seems like it could've worked. But I always like a good major version skip, e.g. Winamp (no 4, because they combined 2 and 3 to make 5), PHP (no 6, because... reasons), and nobody so much as thinks about IPv5.