Thinking about Windows Build numbers

re: Thinking about Windows Build numbers

GregM — Thu, 09 Jul 2009 05:26:00 GMT

It seems very odd to me that the major version number of Windows 7 isn't 7. Any insight on that?

re: Thinking about Windows Build numbers

LarryOsterman — Thu, 09 Jul 2009 05:34:51 GMT

Why is it 6.1 and not 7: http://windowsteamblog.com/blogs/windowsvista/archive/2008/10/14/why-7.aspx

re: Thinking about Windows Build numbers

JP — Thu, 09 Jul 2009 08:51:09 GMT

With Vista getting build number 6000, I suspected that the build numbering system had been updated to reflect the major OS version number. It seems that either this has never been the case or the system has been changed again -- after all, there is now no way Windows 7 can become build 7000 or 6100...

I also wonder where the 1830 in DDK 3790.1830 came from -- do you have any idea?

re: Thinking about Windows Build numbers

LarryOsterman — Thu, 09 Jul 2009 08:57:07 GMT

JP, I have absolutely no idea.

re: Thinking about Windows Build numbers

James Bray — Thu, 09 Jul 2009 15:55:05 GMT

I'm not usually much of a betting man. But it would be really neat to place a bet on the final Windows 7 build number.

I suspect I'd just be greeted with frowns at Ladbrokes though :-)

re: Thinking about Windows Build numbers

piyo — Thu, 09 Jul 2009 16:19:59 GMT

final build number: 7777

Windows 7 (6.1.7777)

that would be epic....

re: Thinking about Windows Build numbers

Mike Dunn — Thu, 09 Jul 2009 19:24:21 GMT

I've assumed all along that it's going to be 7777. Although 7331 (leet backwards) would score higher on the geek scale.

re: Thinking about Windows Build numbers

asf — Thu, 09 Jul 2009 23:21:29 GMT

Win98 will always be the winner with .1998

I'm hoping for .7777, but .7331 would also rawk

re: Thinking about Windows Build numbers

Ver — Thu, 09 Jul 2009 23:49:19 GMT

Also interesting (if I remember correctly):

from NT to XP (and I think 2003 too), Service Packs don't change build number, on Vista it is build 6000, 6001 or 6002 depending on SP.

Any insight ?

re: Thinking about Windows Build numbers

dave — Fri, 10 Jul 2009 00:22:58 GMT

>The Exchange build numbers were rounded up to

> round numbers for major milestones in the product.

In other words, a perfectly good, simple, and useful idea (which adequately fulfilled its purpose to allow engineers to tell one thing from another) got overloaded with some crap requirements of dubious value, and now people waste their time on it.

(Yeah, seen it all before...)

Here where I work, we seem to have got the 'add another even less significant field on the right' disease of version numbers: 1, 1.1, 1.1.1, 1.1.1.1, ....

re: Thinking about Windows Build numbers

dave — Fri, 10 Jul 2009 00:35:03 GMT

> Service Packs don't change build number,

If version 1234 gets released, then the development team carry on developing, they're going to make version 1235, 1236, ... 1300

Now if someone goes back and makes an update to code from build 1234, what build number should they use?

There are two questions here:

1) Service packs don't update everything. A build number is the build number for 'the system'. Do we want to permit a case where each thingy has its own build number?

2) If build numbers are globally unique, as they seem to be (i.e., if there is an 5.1 build 1234, there is no 5.2 build 1234) then the only way to get a new build number is to grab the next free one (unless you preallocated a batch!). Suppose you call your service pack build 1301. Is it considered weird for build 1301 to be back-in-time, overall-functionality-wise. from build 1299?

re: Thinking about Windows Build numbers

LarryOsterman — Fri, 10 Jul 2009 00:57:03 GMT

Dave: That's where the "QFE number" I mentioned above comes in.

re: Thinking about Windows Build numbers

dave — Fri, 10 Jul 2009 01:23:25 GMT

You really need to add another field on the right ;-)

re: Thinking about Windows Build numbers

asf — Fri, 10 Jul 2009 01:32:39 GMT

@Ver: yes, Vista was the first OS where a SP changed the build number AFAIK. It's probably related to the fact that VistaSP1=2008RTM but why, I don't know (Since GetVersionEx already gives you the SP version)

re: Thinking about Windows Build numbers

Maurits — Fri, 10 Jul 2009 18:37:35 GMT

Obvious solution: drop build numbers altogether and use GUIDs.

re: Thinking about Windows Build numbers

Mihai — Fri, 10 Jul 2009 19:11:15 GMT

@Mauritus

GUIDs have a major problem: are not ordered.

With the current version number system you can do some tests and decide if your application can run or not on this system (major version > 6, for instance).

re: Thinking about Windows Build numbers

Mihai — Fri, 10 Jul 2009 19:25:29 GMT

A really cool versioning system is used by Donald Knuth for TeX and METAFONT:

"The latest and best TeX is currently version 3.1415926 (and plain.tex is version 3.141592653); METAFONT is currently version 2.718281 (and plain.mf is version 2.71). My last will and testament for TeX and METAFONT is that their version numbers ultimately become 'pi' and 'e', respectively. At that point they will be completely error-free by definition."

(http://www-cs-faculty.stanford.edu/~knuth/abcde.html)

re: Thinking about Windows Build numbers

Igor Levicki — Fri, 10 Jul 2009 20:29:37 GMT

>>I also wonder where the 1830 in DDK 3790.1830 came from -- do you have any idea?<<

Maybe it was built at 18:30?

re: Thinking about Windows Build numbers

GregM — Sun, 12 Jul 2009 05:56:54 GMT

Larry, thanks. That at least gives an explanation. I still don't think it makes any sense, but then again I'm not the one dealing with the compatibility problems. What kind of compatibility problems does bumping a major version number cause that bumping a minor version number does not? Is it just people doing version checks wrong, and they're less likely to royally mess them up on the minor version?

re: Thinking about Windows Build numbers

asf — Sun, 12 Jul 2009 07:29:38 GMT

6.1 is the "correct" version number (nothing to do with compatibility, Vista kernel was a major change, 7, not so much, same as 2000 to XP)

The stupid marketing people just has to go with Win7 (Along with idiots that don't understand that MS has been doing a major,minor[,minor] release cycle for 10 years now)

The name Windows 7 would have been so much cooler if they had waited so the actual version was also 7.0 (I guess there is always Win8) I guess they had to really separate it from Vista and did not want to go with another weird name. IMHO it should have been Windows 2010 or Windows 20X for that OSX feel, or Windows MMX for that superbowl feel (With a dash of processor nostalgia)

re: Thinking about Windows Build numbers

Aaron Margosis — Sun, 12 Jul 2009 07:52:04 GMT

@GregM - here's an example of a VERY common bug. Someone built a program back in 2002 and wanted to make sure that it ran on Windows XP (5.1) or higher - no Win9x and no Windows 2000. So they coded a check like this:

If Not (vMajor >= 5 AND vMinor >= 1) Then

{

DisplayMessage(“This program requires Windows XP or newer”);

LayDownAndDie;

}

When you run that code on Windows 6.0 or 7.0, the minor-version check fails. This same problem cropped up with programs written for Windows 3.1 when they ran on Windows 95. The solution on Win95 was that any 16-bit program that asked for the version number was told "3.95". 32-bit programs were told 4.0.

re: Thinking about Windows Build numbers

LarryOsterman — Sun, 12 Jul 2009 18:28:50 GMT

asf: Actually it's everything to do with compatibility. There were probably as many changes made to the kernel in Win7 as there were in Vista (major changes to the scheduler to remove the dispatcher spinlock, for example).

re: Thinking about Windows Build numbers

Anonymous Coward — Mon, 13 Jul 2009 03:50:42 GMT

@Mike Dunn, XP's 2600 build number was by far the most geekish build number. I wonder if just happened to be the closest round number or if there was an effort to target 2600 specifically.

re: Thinking about Windows Build numbers

Aaron Margosis — Mon, 13 Jul 2009 18:18:10 GMT

@Anonymous Coward: I had *heard* that it was specifically targeted.

re: Thinking about Windows Build numbers

Craig Barkhouse (MS) — Mon, 13 Jul 2009 22:34:15 GMT

My guess for the final build number is 7600. ;)

re: Thinking about Windows Build numbers

asf — Mon, 13 Jul 2009 22:51:41 GMT

@LarryOsterman: yes, but thats internal to the kernel (Not a whole new driver model and "new" usermode "security" model). from 2000 to XP the "inner loop" and registry access was improved etc. but that does not make it a major release

The compatibility argument is bullshit, sure some idiots do:

if (major>=5 and minor>=1)

but that is not the only way to screw up the version test

besides, its not going to stay 6.x forever

re: Thinking about Windows Build numbers

GregM — Tue, 14 Jul 2009 00:45:20 GMT

"When you run that code on Windows 6.0 or 7.0, the minor-version check fails."

Okay, so this would be compatibility between XP and Win7, not between Vista and Win7. Programs that don't currently work on Vista would start working on Win7 because it is now 6.1.

re: Thinking about Windows Build numbers

steveg — Wed, 22 Jul 2009 07:31:47 GMT

what about the last part of the numbers for win7.7600.16384 and win7.7600.16386 part of the version numbers? eg 0x4000 and 0x4002.

Are they a set of flags indicating anything interesting? Or are they the "real" build number?

re: Thinking about Windows Build numbers

LarryOsterman — Wed, 22 Jul 2009 08:20:25 GMT

steveg: I answered that in the post - the 16384 thingy is the QFE number.

re: Thinking about Windows Build numbers

steveg — Thu, 23 Jul 2009 04:33:07 GMT

> steveg: I answered that in the post - the 16384 thingy is the QFE number.

Ta!

re: Thinking about Windows Build numbers

Charles Oppermann — Wed, 12 Aug 2009 06:31:37 GMT

>>I also wonder where the 1830 in DDK 3790.1830 came from -- do you have any idea?

I believe the .1830 is the time of the build (6:30PM local). Sometimes when a build stops and is restarted, the time of the build becomes necessary. This is because one or more variants of the build migh have completed and propagated, but there will be another build with the same build number coming along.