Mike Bazarewsky: Stuff I want to post

WinFS Beta 1 released!

WinFS Beta 1, including XP support, has been released and is now on MSDN!  (I'm downloading now, wheeee!)

Longhorn/Vista NTFS

There are some significant new features in NTFS in the Longhorn/NTFS time frame:

• NTFS repair is going to be at some level online, meaning that not all corruption will require an off-line CHKDSK.  This is actually pretty cool, and there are a lot of mainstream OS filesystems out there that cannot do this.  This will be a big step forward.  However, this will also tie in nicely with...
• Transactional NTFS... for the entire history of NTFS, NTFS has journaled metadata, to help recovery on NTFS filesystems.  But, this has not helped with data changes, only metadata changes.  Well, in Longhorn, NTFS is now data-level transactional as well.  In fact, quoting from "What is Transactional NTFS?" in MSDN, check this out:
  

  TxF can participate in distributed transactions that the Distributed Transaction Coordinator (DTC) coordinates, which allows you to use TxF for the following:
  * Transactions that span multiple data stores, for example, a single transaction for file and SQL operations
  * Transactions that span multiple computers, for example, a single transaction for file updates on multiple computers
  

  This means that the filesystem is at a level that is virtually unknown in other operating systems today. The MSDN entry only hints at levels of functionality that are available.
• Symbolic links are now available.  This is actually evolutionary for NTFS, not revolutionary. Since Windows NT 3.1, NTFS has supported hard links, primarily because of the POSIX subsystem.  (See for example this NTFS overview.)  In Windows 2000, reparse points were added, which offered a completely open method to add a link on the filesystem that connects to code that interprets that filesystem location.  (This is how HSM, mount points, junctions, and so on work.)  If you think about this, this is also a huge amount of power, much more than say the Unix "devices are files" model, because that model forces you to have a device driver, which is not an appropriate model for a lot of functions.  Anyway, symbolic links are a reasonable addition to this.  I don't know right now if these are implemented through reparse points, but that would be a reasonable guess.  (I need to play with this more under Longhorn.)
• There's some new international support, but to be honest, this does not matter to me in my daily life or to most (all?) of my clients.

So far, these changes appear to be backwards-compatible at some level.  My laptop has XP Pro SP2, Windows 2003 SP1, Longhorn Server x64 Beta 1, and Vista Pro x64 Beta 1 installed, and the XP/2003 installation filesystems (including the shared data filesystems) appear to be fine.

 

Longhorn Server Core

This is not a secret, although some of the details are. On the Longhorn Server side of the current Beta 1 release, there is a Core release of Longhorn Server. Currently the functionality is DHCP, DNS, AD, print server, and file server. Notice this does not include IIS, yet... but here's hoping. Also, something that through me off when I first installed (I installed twice!) is the way the interface works - it's a couple of Command Prompt windows. No Explorer. No Control Panel. It's pretty darn amazing actually for a Windows release. Of course, some people will tell you it's impossible to do this in Windows because you can't do anything from the Command Prompt. Guess they're wrong.

CoolWebSearch - not just evil, but criminal

Wow, this is even more scary than you ever thought.  This is not just "benign" tracking spyware, this is stealing your life out from under you spyware.  Not good at all.  I have cleaned off CWS off of customer machines before (typically when some moron employee decides that installing some crapware is appropriate to do on a chttp://sunbeltblog.blogspot.com/2005/08/massive-identity-theft-ring.htmlompany machine), and it's often quite difficult.
More at the original source, the Sunbelt blog - check out the original post and the follow-ups as they happen.

(Edit 2005-08-09: CoolWebSearch was not actually doing the keylogging, apparently - it was more like there as part of a CWS infestation.  So, CWS is not the primary evil here.  Just a convenient target.)

Windows 2000 Roll-up to be Re-released

Hmm, just the other day we had a customer with serious issues from the Windows 2000 Roll-up, and now I read it's going to be re-released!

IE7, Acid2, etc.

I have been busy with real work (e. g. training and consulting) so much lately that nothing's been here in a little while.  But, there are some very interesting MS-related and tech-related items about right now that I wanted to note.

First, there's the discussion about "DOS ain't done until Lotus won't run".  Slashdot had a story which linked to Adam Barr, a Microsoft blogger, basically destroying that old myth. If you read the comments, you can find reference to DOS 3.31 vs. 3.30, and about horribly broken copy protection and DOS 1.0 vs. 2.0, but in general, nothing that remotely suggest purposeful breaking of Lotus 1-2-3. Speaking for myself, I remember running Excel 2.0 on my old Tandy 1400HD (so old, I can't find a direct picture, only a 1400LT reference, which did not have the hard drive), using the shipped Windows run-time version, after having used Lotus 1-2-3 for a while, and thinking, "wow, Lotus is garbage." Well, I probably didn't exactly think it in those words, but it was pretty clear that Excel was a better product. The market apparently agreed; by the time Lotus had a Windows version, it was over.

Next, there's discussion about IE7 and Acid2. Ars Technica, Extended64 (twice), and Robert McLaws, among others have discussed the Paul Thurrott posting about boycotting IE7 over their statements in the IE blog about not passing Acid2 at release. The folks who invented Acid2 are quite happy with IE7- you can see this for yourself. It appears from reading comments at many of these sites that many people just don't get the Acid2 test. Passing it means you do a selected set of things a certain way. It does NOT mean you are CSS 2.1 compliant. And failing it does NOT mean that you are not CSS 2.1 compliant. Anyone who says otherwise doesn't get it. Period.

Finally, there's a new mouse from Apple. Ars Technica had an early review, and you know what? I say, "Apple, welcome to 1996!". So much for the since-day-one nonsense about "a single button is all anyone needs, anything more is too complicated." Interestingly enough, Apple still thinks its users are total morons, because the new mouse starts with both left and right side buttons doing a left click, at least for the MacOS X out-of-the-box default. (Read the Ars review for more on this.)

Sometimes truth is stranger than fiction.