Slashdot Mirror
Gates Explains Longhorn Delay, Diet
An anonymous reader writes "Microsoft has set late 2006 as the deadline for shipping Longhorn, but to make that date, it had to delay the full implementation of WinFS, an ambitious file system geared at letting users search through all of their files at once. In this interview with Bill Gates, he provides a summary of why Microsoft decided to drop WinFS, saying: "WinFS, I'd be the first to say, is very ambitious. Nobody has ever brought together the world of documents, media and structured information in giving you one simple set of verbs that lets you richly find, move around and replicate those things." Meanwhile, MS Watch has published Longhorn head-honcho Jim Allchin's memo on why some Longhorn features had to be axed."
Press releases like these are free ads for Microsoft. Does anybody here not think that Microsoft knew this was going to get released:
We will not cut corners on product excellence. Our powerful vision is intact; our shipment plan changes will let customers get access to parts of the vision faster.
Why don't they just admit that the market is forcing them to release parts of Longhorn (like Monad) earilier than expected! Leaks of betas and press releases like these are easy ways to keep the Microsoft buzz elevated.
If they didn't release a product until 2008, the market (mostly linux) would have time to catch-up.
So, in his (apocryphous) diary, he mentioned being the inventor of product pre-announcement, now he's just invented the post-pre-announcement. :)
Way to go, Bill
Trolling using another account since 2005.
"Nobody has ever brought together the world of documents, media and structured information in giving you one simple set of verbs that lets you richly find, move around and replicate those things."
Wasn't this the whole idea behind meta-tags for files? I thought thats why we had such tags in windows media too?
Or is this the same tags that winFS will use to search with?
- http://www.milkme.co.uk
"WinFS, I'd be the first to say, is very ambitious. Nobody has ever brought together the world of documents, media and structured information in giving you one simple set of verbs that lets you richly find, move around and replicate those things."
Maybe Bill considered them nobodies...
WinFS, I'd be the first to say, is very ambitious. Nobody has ever brought together the world of documents, media and structured information in giving you one simple set of verbs that lets you richly find, move around and replicate those things.
Translation:
We thought it was a good idea but no-one else has done an implementation that we can copy off, so we can't really figure out how to do it.
Can anyone explain exactly what will be in Longhorn, now that the new filesystem and graphics system is not going to be in it ?
"Free software as in beer, copy protection as in racket" - Telsa Gwynne
So that's bye bye new file system
bye bye new GUI
bye bye new API
wtf is left ?
http://www.theregister.co.uk/2004/08/27/microsoft
There are places where the networks are not touching,and there are places where they are-Boeing's Lori Gunter
One is (that) we have a date-driven release. Things that make that date get in.
Previously Microsoft were skirting around the 2006-7 point without being clear about when Longhorn would ship; it looked like they were going to try to finish features X and Y before release. So now they've moved on to a date-driven release, we can pretty much guarantee 2006 for Longhorn (client edition) and they're going to drop anything they have to, to make that date.
Bill said that the OEMs are okay with the delay, so why the pressure? Looks like Linux is hurrying Microsoft up!
I must admit I'm getting more and more of the deja vu feeling, reading Microsoft's statements on Longhorn. I've seen it before, when Apple representatives struggled to explain the delay with shipping their ultimately sophisticated version of MacOS, codenamed Copland. They understood all too well that the classic MacOS is a bloated unstable construction based on a single-user single-machine Macintosh System, that was not designed with networking and multitasking in mind. They managed somehow to hack this system to have a sort-of poor man's multitasking and also some rudimentary networking capabilities, but they knew it's not gonna last in the Internet Age. They needed a new system and they needed it ASAP. Yet after millions of bucks and years of coding, Copland turned out to be just nothing but very expensive vaporware, and Apple's last chance to survive was to purchase NeXT, with their Unix experience, and thus MacOS X was born.
There are many similarities with Windows and Longhorn - Microsoft also tried for a very long time to hack and upgrade their old OS, also designed for single user with no networking. And yet they were strangled by their own limitations they needed to keep for sake of backwards compatibility. Can they solve it on their own or will they just, say, buy Sun for their OS experience?
I'm guessing this means that they'll be using some implementation of NTFS with longhorn. Could be good news to all those dual-boot people out there that like to be able to access their Windows files from Linux.
Just as they're making some progress with mounting NTFS filesystems under linux, MS changes the FS. Something which surely would cause problems in Linux.
Looks liks we'll be able to keep dual boots with Longhorn after all.
I guess Linux coders copy MS features for the benefit of those who wish to migrate - not to enhance the power and usability of the OS itself. Secondly, these changes would take a few days in Linux (KDE or GNome); not years as with Microsoft.
-
If you keep throwing chairs, one day you'll break windows....
Introducing Microsoft Longhorn Millenium edition!
Preorder now and recieve a copy of Duke Nukem Forever!
If people are waiting til 2006 anyways, Gates would have been smarter to delay Longhorn until WinFS could be totally implemented. If they need more money coming in on the conveyor belt, then they could have just released Windows XP OSR2 - essentially a service pack/ upgraded version people would have to pay for. I seriously doubt I will be paying for a cippled version of Longhorn - especially if its best parts are going to be made available for XP.
Looks like maybe MS should have spent a little more time getting WinFS working instead of tweaking the UI to make it "oh so pretty." Unfortunately, I think MS realizes that a slick (albeit graphics intensive) UI will likely sell more copies to the ignorant masses than an innovation like WinFS.
Nobody has ever brought together the world of documents, media and structured information in giving you one simple set of verbs that lets you richly find, move around and replicate those things.
Someone please call Oracle and tell Larry that Bill says that IFS (The Oracle Internet File System) doesn't exist.
What is iFS?
iFS can manage all content -- which is scattered across PC desktops, document management systems, and websites -- in a single repository, he said. It supports the storage and management of more than 150 different file types, including documents created using XML.
"Nobody has ever brought together the world of documents, media and structured information in giving you one simple set of verbs that lets you richly find, move around and replicate those things."
Didn't BeOS have something similar?
Also, won't OSX actually have something like this even before Longhorn ships (without WinFS).
Aren't there a lot of pretty advanced projects to do the same for Linux, for example beagle for gnome and the new kde search feature planned for the next release? (Granted, these won't be implemented at the fs level, but who cares as long as they work)
Isn't reiserfs4 actually providing some of this functionality (and much more) and has allready been released?
Doesn't MS have about 60 billion Dollars in the bank and still can't get its act together?
Didn't MS talk about something similar already years ago and wanted to ship it with what is now known as Win2000?
Maybe he should have a look at iTunes and GMail.
For me, a kind of "iTunes for files", including smart queries, would be fairly enough. And it doesn't require a brand new file system and its instability risks...
It's not about finding files by filename, but about finding files by content.
Obviously you are trolling but this is a common belief...
However, Monad is obviously a way that Microsoft is trying to catch-up with the powerful scripting ability of *nix shells.
Of couse, some linux installs with have sidebars and other copies of new longhorn features. Longhorn will likely gain some new linux-like features between now and then as well... It's just the features race.
In competitive software markets one product will always try to match the bells and whistles of similiar products. For example, IE gained pop-up blocking.
Talent borrows, genius steals.
AC
Don't be silly. What they're looking at is something like GNOME Storage where you can type in some search terms and semantically find the files.
Something like 1960s music or e-mails to Bruce, I'd guess. WinFS ties up all your documents, media, mails etc. into one database for indexing and searching, and beats the hell out of DIR C: /s/a.
imagine that... treating everything as files...
;-)
how inovative...
So, what we have been shown in the next release of OSX Tiger that lets you search your documents, email and file system isn't anything like this. We have seen it in action and the set release date is 2005.
Come on Bill....Steve can pull this off and he doesn't have 50 billion in the bank.
Evolution or ID?
> Any guesses?
Yes, actually. That you have absolutely no idea what you are talking about. Come on, do you really believe that the windows development team would give that much weight and media time to a system that implemented find / -name $string -print?! And even then, that they couldn't hammer it out in a day? Please.
What they are looking to do is to integrate the filesystem into a database system, where files are organized not by directory, but by use/type/relationship. Even I have a hard time wrapping my head around what this will look like once it's carried out. What will it gain us in user experience? My gut says 'a lot' given the sheer amount of development time these people have put into the project.
I certainly feel anger, fury and loathing when simpletons critique what they don't understand.
winFS doens't seem very usefull in my eyes. it's just a layer on top of ntfs. in the end (windows 2012) you'll see they rewrite it to be a true filesystem. reiser4 seems to do this the right way. having a nice filesystem that you can extend all the time using plugins. I think microsoft wastes a lot of time by doing this in 2 steps. I also understood that winFS is "My Documents" only (or something like that) and cannot be used on the entire harddisk (atleast not in longhorn).
WinFS, I'd be the first to say, is very ambitious. Nobody has ever brought together the world of documents, media and structured information in giving you one simple set of verbs that lets you richly find, move around and replicate those things.
*cough*
Microsoft still can't come up with shit until Apple has done it better, first. Sad.
Always ask 'why?'
If MS did nothing innovative before 2006, it (Microsoft) will have to do the catch-up.
He said "the market", you're talking of "the product". Those two are unfortunately nowhere as closely related as one might wish...
Kjella
Live today, because you never know what tomorrow brings
No. updatedb and slocate find on the filename, not contents.
blah
Nobody except the people who brought you BeOS and Hans reiser has done a filesystem like WinFS :-)
WinFS is a blatant ripoff of the BeOS filesystem.
What Microsoft REALLY needs is a next-gen OS. The current codebase isn't going to hack it. The delays on Longhorn are an absolute giveaway. If Longhorn had come out in 2004, it would already have been out of date. 2006? Don't make me laugh.
Unix-like systems are going to win out in the end. That is why Mac's OS X looks like a smarter move every day.
Microsoft has so much cash and so much clout that it will take a long time to die, but it is doomed to do so unless at some point it ditches backwards compatibility and the current codebase and does something new.
I'm not wrong. You haven't thought about it hard enough.
Does anyone else think WinFS is a Bad Thing? A filesystem is a low-level, simple, reliable method of storing files on a disk and a database is a method of catologuing and searching through files. If you combine them, it will get hideously complicated. Which means it will probably be buggy and slow. It's almost as bad as putting windowing in a kernel...
Don't you hate meta-sigs?
I really don't see what difference it makes as long as longhorn is released in the next 4yrs. No matter how many computer-savvy people decide not to use it, it will still be THE os.
It matters because the market is now aware of Linux, which it never previously was. It has major corporations backing and investing in it (IBM, Novell, HP Compaq, Sun) and it has not only mostly caught up with the "features" of Windows but has surpassed them and approaching the kind of features slated for Longhorn.
Just look at the 6.8 release of the X.org X11 server. With the composite extension and cairo you'll be able to do pretty much anything offered by the Longhorn GDI. Of course, it needs to mature, to be further tested, to be further accelerated, and to have enough applications developed for it to become useful... but I think between now and mid-to-late 2006 is more than enough time for that to happen. Add to that the network transparency of X and all of a sudden Microsoft will be playing catch-up in that respect.
Also, look at Storage and the various other FOSS projects working towards that goal. It looks like WinFS may even be late in that regard to, again playing catch up.
Put all this together with the market momentum Linux is gaining (don't be surprised if it hits double figures in terms of market share by 2006) and Microsoft's position as the dominant OS player will be under massive threat.
Also, they can't afford to fuck up again on this one. The world is getting very impatient with the whole security mess. It's simply costing businesses too much to keep on top of it. FOSS operating systems have a far better security record making them even more attractive.
I could go on and on, but Microsoft is betting their monopoly future on Longhorn. And the free desktop could literally beat it to the punch.
Free Gamer - Free games list and commentary
you mean like spotlight?
I used to have a better sig than this, but I got tired of it
The Register interviewed Dominic and Benoit Schillings a couple of years ago and is a very good read.
Are you local? There's nothing for you here!
So you'll convert?
You will take your hundreds (maybe thousands) of current files and insert meta-data into each and every one so they fit the new "paradigm"? I won't, and my guess is that a whole butt-load of soccer moms won't either.
I personally don't understand the need for the concept. I do my development, writing, gaming, and keep my photography on one computer. I find the current file-system completely satisfactory and sufficient for the job.
The way I work in the physical world is the way I work on my system. I keep everything in organized stacks, in specific locations. "Emails to Bob" are kept, for instance, in MyName/Emails/Bob. Not hard at all.
I see all this meta-tagging as making everyone into data entry clerks, and, personally, I don't need that.
I would entertain someone coming up with really functional reasoning explaining the need for all this.
No, not the type of content, the ACTUAL content. Like searching for "pictures of houses" and the system going away and generating a list of all the jpeg images that are tagged with the "house" keyword.
Other useful examples might be "films starring Tom Hanks" or "music by The Red Hot Chilli Peppers"...
"One problem with Konghorn..."
Oh dear Lord. Don't tell me the KDE team are reimplementing Longhorn.
Free Gamer - Free games list and commentary
"Rapid Development" by Microsoft Press. There's this chapter on Classic Mistakes. To mention a few:
- unrealistic expectations
- wishful thinking
- placing politics over substance
- overly optimistic schedules
- inadequate design
- feature creep
Maybe this company should take some time to read their own publications.
Visit http://ringbreak.dnd.utwente.nl/~mrjb/growingbettersoftware to download your free copy of the book
"Isn't reiserfs4 actually providing some of this functionality (and much more) and has allready been released?"
:)
Yes, it has.
I was just thinking that it would be cheaper, easier, and faster for Microsoft to just license Reiserfs v4. Just the atomic file writes/updates would be worth the effort! And the filesystem supports plugins.
Some people in the Linux community don't think Reiserfs v4 is stable... but I'm willing to bet by 2006 the issue will be settled.
The problem with meta-tags is that they have to get populated somehow. Only the anal fill in meta-data, everyone else either blows it off or takes the defaults.
The real breakthrough happens when the system can decode and parse the file accurately to provide "automagic" meta-data. Otherwise meta-tags are a nice academic exercise that is either ignored or misused in practice.
A house divided against itself cannot stand.
- http://www.sunrizen.com/ It basically does what was taken out of Longhorn- turns the filesystem into a database, and uses that for fast searching. It doesn't have the SQL and real-time queries that BeOS does, but it's hella fast and really cool. I've used it for bug-hunting code, since it searches for text inside documents hella fast. It's much better than MS's shipped in search utility.
click me
Now Longhorn isn't going to be shipped until late 2006. Let's give them the benefit of the doubt and say they'll hit that date (just in time for Xmas!). OK, so that means that they will have been working on this thing for a MINIMUM of 5 years. If there was any release overlap, and I am sure there would have to be, it is probably more like 6 years. WTF have they been doing in Redmond!? You can't tell me that everyone there has been working on XP service packs.
Now I am not discounting the complexity of software and what it takes to release something of this magnitude. But we are talking about the largest and richest software company on the planet! Surely if anyone could do this, it would be..... Hmm. Perhaps what seems to be an advantage is actually a disadvantage in this case. If you look at their OS timeline (I used this one ), it seems that it was usually around 3 years between major instances of their OS lines. Now, that has doubled for some reason? Maybe they had to start over from scratch and are putting some security into this one. (the good kind, not the DRM kind)
I guess we'll just have to wait and see. It's good for me that they are delaying, at least they won't be changing the "corporate standard" again where I work. I really don't care for XP and wish I had 2000 back...
My beliefs do not require that you agree with them.
Why don't you check those out to see how much it will do for the interface. What will MS "invent" next?
Sticking feathers up your butt does not make you a chicken - Tyler Durden
Because file extensions suck, that's why. All the rest of the meta-data abouta file (creation time, owner, author, etc) is in attributes, which should the type be encoded in the name?
.dat file was the film itself. Now, .dat is associated with Notepad on that PC - had I just double-clicked it, it would've opened in Notepad. So I had to right click, choose "Open with...", and select media player.
.dat files with media player, as the vast majority aren't films. If the file type was determined by the contents of the file (or some meta-data other than the name), then I could've just double-clicked and relied on my OS to work out what to do with the file. Sure, it's not difficult to choose something to open it with, but then I'm technically-minded. My parents (and some of my friends) would've been unable to play the disc.
Practical example: I have a couple of VCDs. My daughter wanted to watch one, on the PC (as my gf was watching TV). It didn't auto-play, and no application was associated with VCDs, so I had to try to work out how to play it. In the end, I realised that the ~700MB
I can't associate all
It's official. Most of you are morons.
Is LongHorn delayed bcos MS couldn't implement this simple stuff?
Don't be ridiculous. Windows (since 2000 at least) has had an equivalent to Linux's (s)locate tool. Clearly that's not what this is about, as it already exists!
I can't think of a word to describe this feeling of anger, fury and loathing combined.
Why are you so angry? Are you losing money (or anything at all!) because of the delay? Seriously, if Longwait being delayed and scaled back in scope makes you that angry, you need to sort your priorities out.
It's official. Most of you are morons.
> However, Monad is obviously a way that Microsoft is trying to catch-up with the powerful scripting ability of *nix shells.
I think MS just thought it would be funny to release something that would "have to be" called "Gonad" if it was copied and release in open source! (Hmmm, or maybe Gnunad?!)
Able to do something != designed with it in mind. My car will run through a little sand but that doesn't mean I'm taking it offroading.
"Gold still represents the ultimate form of payment in the world." - Alan Greenspan, 1999
"implement the things that FOSS world can't do" eh? Then you go and talk about filesystems and vector graphics, both of which, at present time, FOSS absolutely trumps MS at. Linux has ext2/3, ReiserFS, Reiser4(which was just released, and has the potential to do everything WinFS will do), Storage(another datastore similar to WinFS). KDE and GNOME are both moving to SVG, and are moving along quite nicely. The X.org X server is implementing loads of new graphics features, and since forking from XFree, they're actually getting done. Also, most of E17's base libraries are mostly done, and implement a lot of features MS is in the process of "inventing."
Sticking feathers up your butt does not make you a chicken - Tyler Durden
"WinFS, I'd be the first to say, is very ambitious. Nobody has ever brought together the world of documents, media and structured information in giving you one simple set of verbs that lets you richly find, move around and replicate those things."
Nobody, Mr. Gates? Apple announced this was going to be a key part of OS X Tiger. It is scheduled to be released this coming year, and they have already implemented it in the preview versions of Tiger that they have made available to developers. By all reports it is working just fine, today, right now.
So please, lay off the "nobody" stuff, mmmk?
...what the grandparent poster was trying to say. At least, I took something different out of it than you apparently did.
It's not so much that FOSS can't implement these ideas. It's that they can't, or at least won't, do so in a way that's pervasive for the whole OS. FOSS can, for example, design a new filesystem or display model, but it can't make all of the apps written for Linux support those things. It especially can't make the apps support it in a consistent and comprehensible way.
Microsoft is capable of saying: This is the way we are going to do things now, and if you are going to make software to run on our OS, that's the way it's going to be. If the Office suite, for example, deals with the new filesystem in a certain way, that becomes the Right Way. Instant industry standard. Any software vendor who deviates from that method is going to be looked at as doing it the wrong way.
FOSS can't compell that kind of compliance. Developers are free to support or not support the work of other developers depending on how much time they want to put in or if they think it's a good idea. If there's a difference in vision, a fork can occur.
Don't get me wrong; I'm not saying the FOSS way of doing things is bad, and I don't think the grandparent poster was either. It's just different. It absolutely has its strengths, but it also has its weaknesses too. Microsoft is, perhaps wisely, choosing to try to push the strengths their model has.
Are they going to ship a Linux distro?
Friends don't help friends install M$ junk.
Lately, I keep running into, Gee the Open Source world used to be cool and interesting. They used to talk tech, but no more. Now it is about gossip! However, "if you look at my other hand" Microsoft has this really cool stuff in their blogs and the likes...
I really wonder if there is not some stealth blogging going on...
Now to address your issues...
1) I read MSDN blogs and it is essentially the same material posted by ten different people. It is quite amazing how "monolithic" independent blogs can be. Scoblizer seems to be the only "oddball"
2) Slashdot has always been about both gossip and tech news.
3) More people use Open Source, hence more news will be about CEO's who give press releases about Open Source.
1. Microsoft announces a new search feature with a layer on top of NTFS called WinFS and will be using MS-SQL Server lite to query the data. Huge bloated solution using technology originally embedded into Office 2003. (Office 2003 installs a mini MS-SQL Service, used with Mail Merge, etc). (I don't know which came first, the chicken or the egg. Microsoft may have announced this ambitious plan after seeing the news about Apple hiring the BeFS developers or they did it first and Apple responded, either way file searching has been itching for a major upgrade industry wide.)
2. Apple hires the BeFS developers and within a year integrates the BeFS metatag system into HFS+. It's extremely fast and it works great. Apple calls it Spotlight and it's available to developers right now in Beta form within the Tiger OS 10.4 beta release. Tiger's been updated a few times already. Expect in first or second quarter of 2005 for gold release. The system works across all file types and can handle indexing the contents of files. There is an API for more advanced metatag insertion and application specific search features and interface. I've seen this system in action and it is truly remarkable. Less then a second to retrieve all sorts of data. Email, AddressBook, keyword search in documents, URL's, Bookmarks, etc., etc., etc. It's so good, why even bother organizing one's data anymore?
- Microsoft forgot a primary engineering philosophy. "Keep It Simple Stupid" - KISS! They simply failed in their initial design of WinFS with MS-SQL Server. They need to scrap it and start over. The primary problems being it's too big and bloated and the potential for bugs is enormous. It's too difficult to build queries. They started with the work done on Office 2003 instead of being more innovative and starting over with a better design.
When XP changed it's search abilities I had endless calls from developers who could no longer search the contents of source code files or SQL files like they could with NT's Find command. Apparently, one had to write a plugin to the MS Search engine to add support for various file types. There were work arounds but they required re-indexing all of the files and it took hours and hours to finally start working. Also it was unpredictable in the way it began a re-index. A new file was not immediately available via search. If Longhorn really does not ship with WinFS then it is deeply disappointing. Well back to giving my developers a grep GUI...
The Apple Spotlight system instantly and on the fly indexes the metadata. It does so very quickly. The results are instantly available. You can save the query and add it to your sidebar so it's available from the main file manager (Finder). Click the smart folder (saved query) and it's always up-to-date with the latest data results. The Smart Folders idea was from iTunes, it's a way to represent a query.
Here's to looking forward to OS X Tiger and future Linux systems using similar metatags! And watching Microsoft fumble the ball and have a thirty yard penalty! Gee, by 2010 MS may actually have a viable search system. Perhaps Google will beat them to it by releasing a Windows file search feature. The Google toolbar and SearchBar are awesome all Google needs to do is add filesytem metatag layer and do the same thing as Apple Spotlight. Heck, I would pay for that solution!
...on Linux. IIRC, the whole point of WinFS is not so much the "find anything anywhere" stuff but that a version of SQL Server was going to be a part of the file system, so that, if I read it right, your receipes can be indexed and catagorized in the context of a rdms instead of folders and such on a "real" filesystem. At the end of the day, NTFS is still doing the actual heavy lifting of saying what block on what platter belongs to what file.
I admit to thinking this was kind of a cool idea...a big information store instead of a bazillion files. The actual implementation, I would think, wouldn't actually be that hard...again, you're not dealing with files per se, but with data.
The *nightmare* is probably in how you're supposed to interact with it. When your whole world is made up of the file/folder/cabinet metaphor, trying to define what an "information store" is, and how a user is going to interact with it in some seamless fashion, must be mind boggling complex because the only way it will work is if you have the relationships correctly set up. Photography cataloging programs do it by giving the user dozens of fields for him or her to fill in, and only on those fields that there is data is it useful to search on.
Back to Linux...I think that implementing this, presumably using a Reiser4 plugin + some RDMS, and then have the correct way to interact with it, would show Microsoft up to no end. "Information at your fingertips" is more likely to get the attention of a PHB than "10,000 node cluster" and anything to show how the Linux community delivered when MS couldn't, is obviously a Good Thing.
BeFS was the FS for BeOS. When introduced in ~1997, it was really extraordinary, with 64-bit addressing allowing file sizes many orders of magnitude larger than competitors (also much larger than physically possible), plus extensive support for metadata. BeOS implemented a great MIME-type system to identify file types using BeFS' metadata support, so the file type was cleanly split away from the file name, unlike the DOS/Windows hack of using the file name extension as a file type identifier. Furthermore, certain BeOS apps used BeFS metadata to allow extremely powerful query operations, including "live queries" that were updated every millisecond or so. BeFS was not really a database FS, but it did incorporate some cool indexing features that allowed database-level performance for certain filesystem operations. The earliest versions of BeOS really did use a true database as the filesystem. This idea was discarded due to excessive performance overhead, and BeFS was created as a compromise.
I have not used ReiserFS 4, but it sounds a lot more ambitious than BeFS. At any rate, the Linux BeFS driver is really a compatibility option that does not provide the same features as using BeFS natively under BeOS. fwiw, I would really love to see someone implement BeOS-like queries for Linux using one of the new metadata-enabled FSes.
So in early 2005 consumers will have a meta data file system, and since Mac OS 10.2 they've had 3d accelerated GUIs... Now if WinFS did get released in longhorn (which it won't be, according to MS.) We'd still be waiting until late 2006, for these features.
I wouldn't place too much emphasis on MS's ability to timeline a product to market. After all windows 95 was meant to have the 3D accelerated GUI, and NT 4 was supposed to have WinFS.
At this rate it'll be 2010 before WinFS sees sunlight.
"Our scheduling and predictability on this project has been better than it was on OS 360. So software has not gotten more complex."
Bill seems to be forgetting that OS/360 was one of the first attempts at anything like a modern OS and whole books have been written about the mistakes that were made in its development. Fred Brooks "the Mythical Man-Month" is largely a result of the lessons learned in its development. What's he saying here? Is he implying Microsoft hasn't learned anything about developing complex software since 1960? As cynical as I sometimes am about the company, I don't believe that... they have put together systems successfully that are far more complex than OS/360.
Remember, OS/360 had to run on hardware that was less powerful than anything any Microsoft operating system all the way back to MS-DOS 1.0 has had to deal with. Features like being able to run a variable number of jobs were restricted to the top-of-the-line models, and most early installations ran it purely in a static batch mode with a fixed number of concurrent jobs.
This is a great soundbite, but it doesn't begin to address the question. The best answer to a question like "Has software just gotten more complicated to write?" is "Yes." I don't know if Microsoft accepts this or not, I have no idea, but if Bill Gates answers a question like that with a red herring like "We're doing better than IBM did on OS/360" I fear they're still in denial. So perhaps the best answer to the next part, "What, if anything, does Microsoft need to do as a company to reflect that reality?", is "therapy".