Slashdot Mirror
Microsoft Uncertain About WinFS for XP
Ant writes "As a follow-up to WinFS to be available in WinXP story from a few days ago, BetaNews reports that Microsoft (MS) stopped short of confirming reports that it plans to back-port its next-generation WinFS file system architecture to Windows XP. MS tells BetaNews it is only evaluating the move while also acknowledging WinFS is still years off. "We are currently evaluating making the WinFS storage subsystem available on this platform and will make the decision based on what is best for customers." a Microsoft spokesperson told BetaNews."
The term 'vaporware' comes to mind...
-- kortex "Not everything that counts can be counted, and not everything that can be counted counts"
But I told you so http://slashdot.org/comments.pl?sid=141645&cid=118 66750
---
Programming is like sex... Make one mistake and support it the rest of your life.
Just like every other product Micro$oft puts out, this one will be put off for another two years. Mark my words, we will not be seeing WinFs for a while.
Get your FREE MAC MINI! CLICK HERE! I will send $20 via paypal to anyone that signs up and completes an offer.
Its in consumer's best interests to force them to upgrade lest they be left behind and forgotten.
Douglas P. Price
They realized it's not cost effective. Nobody is going to spend extra money getting xp with winfs. They'd have to give it away free, and they probably realized it would cost them a pretty penny in developer time to get the thing to work, especially games.
"Piter, too, is dead."
Anyone else think this is Bill's white whale?
I bet they do release it -- better to debug their beta stuff on desktops (that mostly affect home users that can't afford to sue MSFT) than on servers where they might get in trouble (PR wise or lawsuits)
Which means in plain english: It don't work yet and when we do release it you will have to pay to upgrade!
By saying that they're not sure, they are safely leaving the escape route unblocked, in case they fall behind schedule again or whatnot.
How many people with WinXP with FAT32 or NTFS would actually format their drives to get the cool new filesystem? I wouldn't!
It will be available for GNU/Hurd :-).
Engineering is the art of compromise.
I don't think it's vaporware... I think they're having problems getting it all together. What they're trying to achieve is quite an undertaking.
Disclaimer: I do not work for Microsoft, I'm not a Microsoft fan, I don't run Windows XP, and I won't buy their coming OS either.
Microsoft has been working on this for ten years. It is never going to happen. This was supposed to have been in 'Cairo' and has been a listed feature in every development OS since then. It is never going to release.
.technomancer
All it would do is make locating files easier, at least that's pretty much how they were shopping it around. You could do that without adding another layer to the HDD by simply having an element of the OS scan in the background efficiently.
Conversely, though, I wonder if the reason they're starting to back off of WinFS now is because including it would mean that all of those obscure file locations where companies like to hide setup files would be that much easier and faster for people to locate. I've lost count of the number of times I've needed to hunt through hidden folders to find some stupid file to edit or delete. And the search taking 30+ minutes didn't help.
Maybe instead of working on WinFS, they should focus on coming up with an alternative to the registry.
That never stopped them before!
A feeling of having made the same mistake before: Deja Foobar
The WinFS or the Duke Nukem Forever?
If I was in microsoft, I would think that backporting a filesystem wrapper over NTFS is probably a bad idea.
It's hard enough to design this WinFS, much less change all the OS components to be compatible with this filesystem. I also think the learning curve/'WTF is this' factor is too great to drop onto Windows XP users. Let it ride on Longhorn, but make sure you give a really full explanation on how to use this meta-data FS well.
I certainly don't find a need for a DB-based FS, but I know that it helps. Will it help enough people enough to make it worth implementing?
Admittedly the feature was little more than a glorified hashtable, but still:
1. This is MS, the company that takes 15 years to realize viruses might be a problem.
2. If it wasn't going to make it in a new version of windows, which accounts for half of all their revenue, how were they going to give it away for free??
Y'all need to stop drinking that kool aid, or at least stop taking it rectally..
Care about electronic freedom? Consider donating to the EFF!
slap a B+Tree index on a journaling FS, and you have better performance than dealing a full transaction-oriented database-derived file system...
if conceptually WinFS is such a good idea (regardless of how MS implements it), open source FS would've already appeared for Linux or BSD....hmmm....
That article contains a wonderful example of the difference between Microsoft and the OSS movement. Microsoft is developing a new filesystem that (one would hope) is vastly more advanced than the one they currently use. Yet they're hedging about making it available for older systems, because they have yet to decide what is "best for customers".
Now, if they were really interested in what's best for customers, you'd think they'd let the customer decide on a case-by-case basis. They could just release the filesystem for older systems via an extensive patch and see what the customers decide to do. Instead, Microsoft is going to determine what is best for all their customers.
The OSS folks would just release (and have released) new filesystems and let the bits fall where they may.
Central planning versus individual choices. Remind you of any 20th-century struggles?
With reasonable men I will reason; with humane men I will plead; but to tyrants I will give no quarter. -- William Lloyd
2. If it wasn't going to make it in a new version of windows, which accounts for half of all their revenue, how were they going to give it away for free??
3. the company which unleashed activex
Remember when the quicken episode, back in '95?
"and just like this i could transfer their balance to my account..."
A feeling of having made the same mistake before: Deja Foobar
From the posting, my emphasis:
...and please order those Microsoft WinVoting units, I don't want to do any thinking come the begining of November anymore.
"We are currently evaluating making the WinFS storage subsystem available on this platform and [We, at Microsoft] will make the decision based on what is best for customers." a Microsoft spokesperson told BetaNews.
Thank you...
Get your Unix fortune now!
A Public Relations representitive from Microsoft was found dead today, apparently tossed from the third floor window at Microsoft Headquarters. According to bystanders, the words "see if this teaches you to leak our stuff!" were shouted after the person was thrown out the window. News at eleven.
It's better to vote for what you want and not get it than to vote for what you don't want and get it.
- E. Debs
>>what is best for customers... any time i hear a microsoft spoksmen say that, i laugh my ass off jordan
I still don't give a flying f%@& about WinFS.
Would you, could you, on a boat?
Would you, could you, while afloat?
I implore you to consider, a f%@& on the sea,
Please care about WinFS, so you do so, AC!
Didn't Microsoft announce a few months ago that they were uncertain about including the WinFS architecture in the OS? Personally...I don't care, the less I have to rearchitecturize (word?) my HDD the more at piece I will be
~Ilyanep
To get message, take amount of carrier pigeons at each stage mod 2. Then decode binary.
... oh wait, no I don't.
While MS is all on about how much better they are or going to be, the fact of the matter is that you'll get there, where YOU want to go, sooner by going more direct and without incompatable file formats, DRM type of constraints, etc..
How often does a company use a cracked version of some sofware package that they actually purchased, so to avoid the problems of the additional protection complexity?
Haha! It's still great vaporware. And a good test of whether someone will suck up any BS Microsoft spews.
--
make install -not war
WinFS also adds the ability for any program to use a supported format through the WinFS API, as the API uses meta data to describe the format to the program, as XML does. This allows the application that created the file to be used through WinFS to access the data, sort of like making applications into libraries. It is similar to piping in Linux, as the program produces intelligable information that another program can make use of.
.sig: Open Source, Open Mind
Database FS's make things easier to search for... but that can be a bad thing too.
If someone found an exploit to run queries on that database, then you can surely find passwords, addresses, vital documents, etc. in a snap!
At least when you obfuscate your folders, you make it harder for both you and intruders to find your info.
Am I the only one that thinks that journalism is going to the shitter. BetaNews posted an UNCONFIRMED story. As in, didn't research it fully.
God it's good to find good news these days. Especially with government "regulations" (aka. Filtering)
DarkMantle I been bored, so I started a blog.
If you want a better way to find your files now, just use Google's free desktop search tool, or do a better job of keeping your files organized. I prefer the latter.
Yes but if i reformat my harddrive, will it be compatible with Duke Nukem Forever? Whoa... slashdot cliches... i feel so dirty. I gotta go take a shower y'all.
At risk of sounding completely stupid, could someone explain what the intricacy is that makes developing a file system such as WinFS such a consuming task? I have limited experience in terms of in terms of developing suite-style applications, but can't comprehend what would make the development cycle for something like this take so long.
Is the coding that difficult? Is getting the standards correct? Testing perhaps? I have no idea.
I've seen many comparisons between Tiger's Spotlight, and MS's WinFS in features... and heard from people I know that have Tiger betas running as full time desktops who say the spotlight problems & updates seem to be where the most work is going into Tiger at the moment.
Does anyone know just what the differences are in concepts here? Is Spotlight going to offer much the same functionality from the point of view of a user? Is it really even the 'killer app' it's supposed to be?
I'm curious as I've heard so much mentioned about it these last few years (10 now with Windows).
Bill: Is it ready yet?
Henchman #1: No, not ye--
Bill: HURRY IT UP DAMN YOU!
Henchman #1 and #2: Yes sir.
Bill: Damnit how long does it take to download the Tiger beta!?
*Bill scowls while looking over Apple's website*
Bill: We must hurry, it will take us surely a year to figure out how to create WinFS.
Isn't Longhorn (with WinFS) supposed to ship next year or so? With WinFS still years off, does it means that Longhorn (and WinFS) is still in vaporware status?
There are >limitless< posibilites for why microsoft would want to try out a new file system, not that I know them all. Most of the time when I read slashdot comments people leave little bits of ideas and information in their comments, kinda like everyones working in paralel out of ram (actually sometimes comments are so off the wall it sounds like they're working out of just the cache) trying to figure something out, but when theres these slashdot posts that have anything to do with microsoft everyone only sees one picture, one view, only one *real* motivation for microsoft to be doing this, lock out linux, duh, totally obvious, _everyone post this fact_
Yes, it could be true, but thats not the point, point is that nobodys is being the devils advocate here and looking for another side.
You guys are using the collective intelligence of slashdot to merely diss something instead of even thinking about it!
Usually whats best for customers is the same as what's best for business, but not in every case. It's bad business to give away to XP users the only thing that makes longhorn worthwhile...
What will be our reason to buy longhorn? And wasn't it announced on here a while back that they scraped winFS for longhorn? Why yes it was...
or else!
If Microsoft back-ports WinFS to 64-bit XP, it could hurt or help them. In one way, it could get more people to go for 64-bit systems. Those same people could make an easy transition to Longhorn because they would already have 64-bit systems.
But in another way, if they go to XP64, they might not have as much of an incentive to go to Longhorn. There would already be one 64-bit OS with WinFS. People might feel that Longhorn is unnecessary.
With all the delays for Longhorn, I wonder if Microsoft fans don't feel like Apple fans during the late 90's, eternally waiting for Copland. During the wait for Copland, Microsoft was basically ahead of Apple, since it already had a true preemptively multitasking OS and Apple fans had to put up with cooperative multitasking and frequent crashes. Now, while Apple is poised to ship OS 10.4 Tiger with Spotlight (aka all the functionality of WinFS) and CoreImage (aka all the functionality of Avalon) before July, Microsoft faces delay after delay. Of course, Microsoft OSes are frequently late (who can forget the many delays of Windows 95?), but though the release came fast and furious for 98, ME, 2000, and XP, Microsoft has been stagnating since then. Even a simple service pack has turned into a huge production for MS to produce and ship.
I think all of these signs point to MS's code base being too big and unwieldy. I don't think anyone doubts that IE is too bloated to fix. Just compare the time between the release of 5 and the release 6 to the time between the release of 6 and now. If Microsoft could implement full CSS selector support and non-broken PNG display, I'm sure they would have by now, but IE is just too tangled to fix quickly anymore.
So, if MS is wandering in a Copland-esque desert, what's to be done? As unbelievable as a suggestion as it may seem, maybe they should take the OS X route and just buy a competitor and cut their loses. Starting over from (not quite) scratch will give Windows a shot in the arm. WINE has already proven that backwards compatibility with Windows applications doesn't have to be dependent on using their existing OS code. They should just buy out Be (a good choice since they already have a metadata filesystem) or someone else with a Unix-like underpinning, and rewrite Windows the right way. It will take another 3 or 4 years, but at this rate, they're going to need that much time anyway. Spinning their wheels on Longhorn won't get MS anywhere. If MS wants to innovate (and that's a reasonable question), it's time to take a chance, kill Copland, and try something new.
Any plans to port ReiserFS to Windows? There are 3rd party tools like Paragon's Mount Everything that allow read/write to ext2 partitions, so it seems Windows can be extended in this way.
Tough day? How about a free Mac mini?
I suppose that there is the usual chaos at Microsoft in the marketing department where a makreting person says something that is meant as the usual Microsoft vapourware in order to gather customer interest but where it is so obviously out of sync with actual developments that someone else has to clarify things a few days later.
.Net by default, Avalon, Indigo, WinFS) to be used by a critical mass of developers and users or else it could very possibly fail as badly as MS Passport did.
I presume that marketing also realised that too much talk about Longhorn features being backported to XP could significantly harm sales of Longhorn when it eventually does come out as people will obviously then simply use those features in XP instead of upgrading, thereby making the usual Windows version chaos (some 15% of all Windows users are still using Win98) even worse and pulling down MS' revenues.
On the other hand, MS knows that it needs to have some way to get the new stuff (XAML,
Damned if they do and damned if they don't. Strangely, I feel no pity with them whatsoever, as it was their own predatory monopoly practices, where they would kill their foes with beneath the belt tactics in order to get that very last 3% of users that they didn't already have, i.e. they were never prepared to sacrifice anything in order to have a cleaner and more unified user base.
Count me out for being a guinea pig and upgrading...
It's coming shipped bundled with Duke Nukem Forever!
"...will make the decision based on what is best for customers."
That's why I love 'em. Always thinking of what's best for their customers.
Insert witty sig here.
Speaking of unlikely partners, I think I would rather Reiser4 be available for Windows.
Karma: It's all a bunch of tree-huggin' hippy crap!
How about creating a Windows file system that is resistent to fragmentation?
Hell, Novell managed to do it ages ago, with the file system that they created for NetWare... 'way back when NetWare 286 was state of the art for Network Operating Systems...
Later, it just got better, with sub-block allocation, as an example.
NTFS is up to, what, version 5? And, Microsoft still hasn't managed to make it efficient... file system fragmentation over time pretty much creates the "need" to replace computers: The defragger that comes with Windows XP, for example, is woefully inefficient... and the users don't run it anyway.
So, over time the perception is that "the computer is too slow"... or, "the server is slow"... when the reality is that, barring hardware malfunction, the processor will run at its rated speed forever.. as will memory... the slowdown comes from filesystem access, which, using NTFS, will degrade over time, if the filesystem is not defragmented.
And,if you use the Windows defrag utility, it won't fully defragment the filesystem: It is a subset of Executive Software's Diskeeper, and so, it's in the latter's best interest to be sure that it doesn't.
As one example: It cannot defragment NTFS' Master File Table (MFT). Another: It requires multiple passes to come close to anything approaching what the purchased version does, and again, who's going to do that?
I wonder how many computer hardware upgrades have been driven by this over the years... more importantly, Windows Server upgrades?
More and more it looks right.
I picture Gates in a cape and tights running endlessly from group to group giving 'genius' guidance to move things along. He finishes and leaves and the people gather and shake their heads.
Cake or Death? Cake Please!
...and will make the decision based on what is best for customers.
HAHAHA! If I had just a penny for every time I heard that from any coorperation, I'd be a millionare! OTOH, if I lost a penny for every time it wasn't best for me, I'd be broke!
...Had this been an actual emergency, we would have fled in terror, and you would not have been informed.
"Microsoft Uncertain About IE7 for XP" :-)
There are an awful lot of /. posts about how WinFS is most likely vaporware (which it very well could be), to about how late WinFS is going to be (which is true), to some odd post about how using WinFS will require reformating your HD (which is unlikely, looking at past M$ options like the convert command to change file sytems and the fact that WinFS appears to sit on top of NTFS).
/. community and asking them to put their code were their blogs are. My motivation is many because I think this kind of file system on Linux a few years ahead of M$ would be awesome.
Anyway, my point is that while I see all the critical posts of WinFS, what I do not see are posts to build a metadata or relational/object based file system for Linux (dare I say LinFS or WINWinFS as in 'WINWinFS is not WinFS').
I guess I am tossing down the gaunlet down to the
I can do QA testing and draft documentation on the project if you'd like. This could be a high profile example of where OSS can succeed (or god forbid fail). It is also a chance to test ourselves and see if we can met deadlines better then M$ or if we run into similar setbacks. It also is a cool way to learn about open source, relational datases, file systems, the linux kernel and countless other interesting core components of computing today. Think about it.
Respect the Constitution
A Microsoft spokesperson also proudly announced that WinFS would be the underlying filesystem for Infinium Labs Phantom Gaming Console. "It's a great opportunity", he stated. "With the optimizations we've added to WinFS, Duke Nukem Forever will absolutely scream on the Phantom."
Xenon, where's my money? -Borno
If you have enough technical know-how you should be able to manage where your files are on the system, at least to the degree that you can get things done. Has anyone considered the possibility that this feature may not be very practical from a user standpoint? What is this really but a faux scheme that allows you to pretend your files are "where you want them"? I can imagine this new implementation to be exploited quite expediently by virus writers and other malicious attackers. Soon, no one will know where there files are on a Windows machine. This may seem skeptical, but I have a feeling that this is just another feature that will be "turned off", if possible, by people who know what they are doing.
They could buy VMS. They would get stability, security, scalibility, and clusterability unmatched by any other operating system. HP would love to sell it because they are floundering anyway. It might take a year or so to port the gui portion windows on top of VMS but that's it.
I am not saying they would do it, just that they could.
evil is as evil does
...and a new copy protection...
We will be seeing popup boxes with something like this : "The file you try to copy contain data you don't have a license for... File deleted !"...
Interesting, especially when Microsoft uses the entire gcc toolchain in their "Windows Services for Unix" thingy.
Hail Hypocrisy.
Hail Microsoft.
Heh. They'd go full circle.
PR stunt gone wrong?
I don't want my file system to take YEARS to conceive and implement. I can't see that leading to anything but it being slow, buggy, and overcomplicated. Better yet why not unleash the win9x shell team to build the next file system and see how quickly MS can nosedive itself.
Sometimes at night I imagine the darkness is filled with horrible things with too many teeth, like Julia Roberts.
Based on what is best for customers, my butt.
They will make the decision based on what's best for Microsoft. I don't think the customer has mattered to Microsoft much since about Windows 95. In fact, 10 years later, I'd argue that customer welfare is near the bottom of their priority list.
Offhand, I can't think of a single move they've made in the last 10 years that really and truly had customers in mind. Being in a monopoly position, their mindset has shifted away from 'what services can we offer in exchange for money' to 'how many feathers can we pluck from the goose with the minimum amount of squawking'.
They've always been nasty, hardball competitors, but at one time they shipped some pretty kickass software, too. Word for Windows was particularly good. Even that horrible flop, Bob, was at least well-intended. But now that they are in a position of real power... if you'll notice, they never, ever ship anything that's really disruptive of or threatening to their main monopoly.
Most likely, their internal studies will be focused around how much money they can make and how much customer lock-in they can manage. Will giving it away free give them enough power to be worth losing the cash from selling it? Should they sell it at a low price, to generate some cash but get it into fairly widespread circulation? Should they sell it at a high price to corporations, to gather lots of cash but gain little leverage over filesystem standards? Should they bundle it only into Longhorn to help 'encourage' upgrades? You can rest assured, thoughts like "Is this technology something that every Microsoft customer should be able to use?" will never even occur to them.
Whatever their actual thought process ends up being, actual customer welfare will not enter into it.
Can you say DrDos?
This FUD is so old it makes Rip Van Winkle look like a 18 year old stud.
First, WinFS winds up not being a new file system, but a system on top of the NTFS file system. (Or, at least, that's their current statement on what it is)
/. regarding MS beating all sides of the horse until it is pulp? What exactly are you looking for? Someone to say that WinFS is going to be great? It isn't, and everyone here knows that MS is wholly incapable of delivering something like WinFS is claimed to be. It'd be like arguing that you could go back in time and buy Manhattan for a few beads + 1.
Second, WinFS is stated to be a DB like layer of the file system, improving search and visual representation by offering multiple views. You'd think they would have done this with email clients first, yet they cannot even make this happen in an intuitive way. I seriously doubt that WinFS will happen anytime soon. Earlier, I'd made the statement that WinFS would indefinitely delay Longhorn, which apparently I was correct in, as WinFS was pulled from Longhorn.
I laughed when they stated it would be released for XP in castrated form - no network connectivity, hell, MS can't even show a "network neighborhood" in a reasonable amount of time for a small network (only about 1000 nodes, takes more than a minute easily, long enough for me to forget to time it or try it again).
In any case, Longhorn will wind up being mostly eye candy that will lead to a slew of new problems (MS "innovates" new bugs like no one else;)
As for Devil's Advocate, isn't anything on
The cesspool just got a check and balance.
I really hate to sound like a troll (because I'm not), but why does anyone even care about Microsoft anymore. What makes people care about the half-arsed 'features' Microsoft tries to wax lyrical about in it's death-throws. They're going down. It sounds like a troll to say this, but I honestly believe in maybe 5 years, 10 at the outset, Microsoft will sell Office, and that's about it. All the techheads/geeks will still have Linux. It will probably have the majority of the desktop, since HP, Dell and all the other big suppliers will start distributing it. For everyone else, who want a computer that "Just Works®", Apple will be there. I see Apple making some fairly large gains back into the marketplace in the next few years, if they play their cards right. Linux too. But Microsoft? Nah. They're dead. Longhorn needed to be out by this summer to give hem a fighting chance. They've blown it, and I think they know it. I doubt Longhorn will ever really see the light of day. I think there's a very good chance that XP will be the last widely-used microsoft OS. In 20 years time, they'll call the last years of the 00's "The Return of the Apple". Although I don't know who's more evil, Darth Bill, or his Steveness!
The truth shall always be free: Boris Floricic is Tron.
"and will make the decision based on what is best for customers"
Uh, don't the customers get to decide what is best for themselves any more?
And give him a cigar too! This is the most insightful thing that I've read on /. this week!
Be careful, playing devil's advocate is often confused with trolling.
I think that the information system gains from having a database-focused file system in Windows XP would be hampered by the need to alter the user interface massively to cope with it. If the way you interact with your documents and storage changes, then the way you interact with the computer to find those things should change too. It doesn't seem the right move; a more suitable move would be Windows Server 2003 with WinFS if that assists the serving it does.
Should the computing power needed for Longhorn alienates a large number of users (as it threatens to do), then backporting WinFS will be an advantage, but one which will need to have a financial benefit to Microsoft (a paid upgrade?). The most reasonable place to deploy WinFS will be to claim that it enhances the security of WindowsXP by blocking access to certain files by software of a particular access level (perhaps in the manner of SELinux), but this seems to me to be more appropriate as a kernel-level adaption, which should be due in Longhorn.
What would WinFS-enhanced Explorer look like? It wouldn't revolve around the Start Menu but around the user space of recently edited files, and with buttons for "New File in Program Foo". You may need a task bar to switch between running applications, but there would be no need to have a Start Menu that's as it is today -- a presentation space for programs you regularly use -- but it would have to become a list of programs and files recently accessed, organised in a way you have specified.
And I went to your schools...I went to your churches...I went your intitutional learning facilities@!
send me your money fachizal my nihzihl!! ...(damn, that looks hebrew)
Did you actually read his post?
He said that adding a database layer over an existing FS was a bad idea. If I understand correctly, BeFS was, from the core out, a database-like FS. That's a completely different thing.
If that were what MS were trying to do, that would be a good idea; they'd have to provide a compatibility layer to make it look like the old (current) style of FS, but I can't imagine that could be anywhere near as bad as trying to stick a DB layer on a non-DB FS.
However, since I'm about as rabidly anti-MS as they come, I'm always happy when I see them coming up with more bad ideas ;-)
Dan Aris
Fun. Free. Online. RPG. BattleMaster.
(Hour later: Okay, it's up now) Cindy: Why is my computer so slow and the disk light on all the time? Programmer: It's indexing every file. Which requires unzipping every zip file and cab archive and calling upon special document translators to extract the text therefrom. It all goes into a big hash table in RAM or VM more likely. Cindy: Why is it really slow for about a day after I install this other app? Programmer: Well, it had to index all the installed files, including all the help files that are already indexed by the help file system. But don't worry, you can set a checkbox for "low priority indexing". Cindy: So this gloabl index of everything may be hours of days out of date? Programmer: Wellllll, yep. Cindy: Hmm, maybe it's not quite ready for the average Joe yet?
It will be faster, allow plugins, and is available right now.
Virtualization rocks. One thing I would like to see, though, is a way to have the windows opened by an application in a VM in their own windows in the host OS, the way some X servers can use the host OS' window manager to open their windows, or a Citrix-hosted app that just gives you the app, and not the desktop.
This might make the VM-hosted apps slightly more responsive (since you're not dealing with the overhead of the guest OS display), and make it seem more native when running applications out of a VM.
Hmm - I'm not sure if I'd actually choose to do it that way. For me, the WinFS stated metadata aims bring up more questions than they solve, at least, looking at this intro. They describe the following scenario:
:)
One reason people have difficulty finding information on their computer is because of the limited ability for the user to organize data. The present file system support for folders and files worked well originally because it was a familiar paradigm to most people and the number of files was relatively small. However, it doesn't easily allow you to store an image of your coworker Bob playing softball at the 2007 company picnic at a local park and later find the image...
Obviously, they're right that filesystems do not provide a lot of support for this sort of thing. But it is not an entirely unsolved problem; you can for example use a photogallery program to store your images, which brings in metadata and a DB right there. If the system you are using is smart, it will make use of something like EXIF metadata anyway and store metadata within the actual image files to the greatest extent possible.
It's my opinion that metadata, short of a radical shift in computing standards, belongs in data files rather than in the OS. Otherwise, sending that file by email (short of a shift in the way email works - and yes I know the promotional literature says that the metadata will be intact in the file when exported to NTFS) will dump the metadata and leave you with an image with zero available metadata. There is maybe a good argument that there's a place for a unique ID for data rather than files, and for a 'link resolver' type functionality to avoid breaking links when reorganising the disk....
This WinFS thing seems to me to be a bit like the fabulous Windows registry; putting centralisation in where it does not have to be. It sounds like a great attempt from the documentation, but I really think there's a more UNIXy way of achieving the same results, without monkeying around with the filesystem. Metadata-enabled locate
Plus, that 'company picnic' example suggests that they haven't really thought about the most major problem in the metadata world - how to get the metadata in the first place. It's as though they think that merely adding the theoretical capability to store metadata will suddenly turn us all into competent archivists.
Somehow, this seems like Microsoft is dangling a carrot in front of us all, then when we get exicted about it it moves out of reach again. (Oh, look! WinFS, the "next-generation" filesystem is coming to the "embattled" os Windows XP...) Yeah, Linux is looking better after this game.
Windows has detected an undetectable error.
Dear Bill:
What we need is a SECURE, STABLE OS that protects users from malware (by DESIGN). No more activex crap, no more "browser included!" bloats. Stop adding new features and give us a "WinXP done right".
WinXP would work pretty well without all these spyware/malware/viruses stuff. Bill, go ask Linus, he knows the secret.
(And please, Bill, stop this "world domination" idea you got. It's not healthy)
NT4's defining characteristic was the switch away from an interface that resembled Windows 3.x to one that resembles Windows 95. It came out in July 1996.
Is this a sigs-optional kind of place? 'Cause I am totally down with that if you know what I mean.
- That comment about centralization where it does not need to be and how that could result in data lose when files are sent elsewhere off the file system.
- The comment about how do you get the metadata in the first place. It does seem far fetched to assume that the user who cannot find their pictures within their present computer will be skilled/focused/organized/accurate enough to properly setup metadata for those pictures.
- You hint at methods outside of the OS, like encoding the data in the file itself. I am going to google "EXIF metadata" after this post, but presently I think of things like XML and relation DBs. What exactly do you have in mind? Any links you could provide would be helpful.
- EricRespect the Constitution
what makes WinFS better than any of the other desktop search engines now available?
Google Desktop Search
Copernic Desktop Search
X1
DocYouMeant Hound
Find Files XP
or any of the other hundreds of file search programs on Tucows. From all of the descriptions I've read, it seems totally pie in the sky. One commenter said you would go to the games library and then click on quake 3 instead of going to c:\games\quake3. Who does that anyway? You go to Start->id Software->Quake or Start->Games->Quake if you've organized them. However, if nothing else, it will at least give Intel/AMD a reason to keep making faster processors. I just hope Microsoft doesn't monopolize the desktop search space in the process.
After all, they back-ported FAT32 to Win95a, didn't they?
Oh, wait, what's the reason we divide things into 95a and 95b again?
Forget it, Microsoft wants to tax us again and there's little reason to believe that they won't use every trick up their sleeves to get us to buy into yet another iterative, evolutionary "upgrade." NTFS may not be all that great, but compared to spending another $300, it's pretty damned good.
You hint at methods outside of the OS, like encoding the data in the file itself. I am going to google "EXIF metadata" after this post, but presently I think of things like XML and relation DBs. What exactly do you have in mind?
Various specific file formats have their own methods of encoding metadata, usually in fairly arbitrary ways (which can be a problem). For example, MP3s have their ID3 tags either at the start or end of the file, EXIF has photo-specific details, DOCs have all that microsoft authorship stuff. A great improvement might be if we could all fix on a standard for the actual metadata to store (Dublin Core, say) so we don't end up with completely different standards in each file format and lots of mapping to do.
There are sort of 'wrapper' XML formats, like METS, which are extremely handy for providing files with some sort of context (here's the official site). As it says on the site, The METS schema is a standard for encoding descriptive, administrative, and structural metadata regarding objects within a digital library, expressed using XML.
Re: the second point, I think maybe people get carried away, as in the MS WinFS documentation, when they think about the information available in an image. There is some easy to collect information, depending on the source of a data file: if it's an image, maybe you know whose camera it comes from and when, maybe you can even guess what that person was doing or where they were when the image was taken; if it's from the Internet, you know everything about the process the user went through to find that file in order to download it (I'm looking at that one atm). I'd be tempted to look in great detail at the kinds of information that might be useful, and set up open standards for encoding and transmitting that information, not to mention gaining an idea of the privacy implications, before going for any filesystem embedding of it.
If you're interested in discussing this further I'm kaiidth at altern dot org.
I agree...? Metadata shuld be extracted from the file been indexed, no extra-info shuld be used. Once the metadata-fs is available, aplications could be modified to add metadata info to the documents.. Maybe the same "method" or api used to extract metadata info from documents could be use to add it or modify it (always into the document).
If enough of the API specifications and functional description for WinFS are published, some young folks with lots of time on their hands could probably write a WinFS work-alike and have it ready before Microsoft.
Wouldn't it be insane if Linux supported WinFS before Longhorn or XP?
What better way to lock users (and even developers) into a platform than to remove the concept of how and where their files are stored? Tell them it's in there, sure. Tell them they can sync it with their msPod or their WinPDA, sure. Tell them their Visual C++ project is grouped into Headers and Source code, no problem. But tell them that their C++ code could be portable, and that they could copy it to OS X or Linux if they could only find the actual file? No way.
It's on my hard drive somewhere... if only I could find it... crap...
I mod down pyramid schemes in sigs.
If you think like a Microsoftie you see how this works. They have the Company picnic scheduled in Outlook (time, date and location), and have exchanged a few messages with Bob telling him they would meet him there, and messages with the organiser asking what activities will be taking place, (softball).
When the download the pictures off their camera, the system matches the available metadata from the pictures, with the rest of the data available from other sources. The pictures were taken at the time the company picnic was occuring and they said they would be attending, therefore the pictures were taken at the company picnic.
It might ask them to identify who is in the pictures (drag and drop into smart folders representing contacts), or it may run them through a face recognition system matching people with the list of picnic attendees.
Even so, if they do a search for "Bob playing softball", it will be able to find both Bob and Softball mentioned in emails relating to the Picnic event, and a set of photos taken at the time.
Its not a matter of manually tagging everything. The more useful informtion you put into WinFS, the more relationships it can create to help you locate things.
"Taligent is still pure vapor. Maybe they'll be the last who jumps up on Openstep... "
Yeah... I do agree that you could work out a lot simply using data from Bob's calendar. Assuming of course that Bob was one of those people who is fairly well organised about his schedule. I'd be surprised if face recognition was much of a success in this context, though you never know I suppose - but I imagine you could get moderate success looking at photo features.
I'm just not too in love with the total lack of external tech specs and whatnot. I have the feeling that this is designed as one of those things that Just Works (TM) provided you're using nothing but the newest MS stuff, but which will abruptly disappear in its entirety as soon as you leave their APIs. You don't need WinFS to store metadata about photos - you could do exactly what you just described in a simpler and more portable way without tying it into a filesystem as such. WinFS doesn't actually help you at all in gathering the metadata, that's a separate problem; it's just a storage medium for it plus a query service.
It could as well run as a relations query service and a resolver service, based on harvesting of data embedded into files. Certainly there are disadvantages involved, since unless you had an OS hook updating segments of the db when you added/modified files, it would not update in realtime, but there is the advantage (which is fairly massive, at least to me) that you're not looking at any fundamental changes, you're using existing technology to achieve an effect. The alterations required are relatively minor.
Well the main point is that it becomes part of the OS platform, rather than just a database in a particular application. I used Outlook in my example, but Thunderbird and Sunbird would work just as well, provided they used the underlying system concepts of Contact and Event.
This means you are less dependant on all the latest MS stuff, so long as the applications stick to the system provided schemas.
WinFS acts as the common platform that allows anyone writing a photo management application to interpret metadata from all the various sources available, without coding for each one individually. Should Picasa have to know how to look inside Eudora's contact database?
Part of WinFS is about harvesting the information already embedded into files, and keeping the embedded information in sync with the database relationships.
It hooks into the file system calls so that it is updated in realtime, ie. when tou use Winamp 2 to modify a ID3 tag through a Win32 file api, the database catches the write and updates its record of the metadata.
Using the shell to change the same attribute would go through the database API, but the engine would then also update the ID3 tag in the file stream.
"Taligent is still pure vapor. Maybe they'll be the last who jumps up on Openstep... "
Well the main point is that it becomes part of the OS platform, rather than just a database in a particular application.
That's what I dislike about it.
The minute you move away from your own home PC, you lose all that metadata. That's what I mean by 'leaving their APIs'. Unless you postulate a whole bunch of altered protocols, etc, in order to make this stuff transferable.
And that is why I would try to keep metadata within data files wherever possible, and make use of open standards in order to encode that metadata for storage or transmission where it is not possible. I don't see that the filesystem is a particularly good place to store it. There are existing solutions to this sort of problem, which are certainly less grandiose than an entirely new filesystem, but have the slight advantage of being standard.
As for your last point, yes, no matter where the metadata is stored, the filesystem will have to be hooked into it in some way. But that doesn't mean that the filesystem actually has to be the guilty party - it just has to call it.
Well any kind of fast searching and browsing will require the metadata to be read from all the files and placed into some kind of indexed structure.
You don't want to be doing a query that requires thousands of individual files metadata to be read from locations all over the disk, there has to be a central storage, and you need to use an API provided by the central database to peform queries.
The remaining choice is whether you access and write the files though the central storage API or the standard platform API. Apple is going one way, Microsoft the other, though for backwards compatibility both methods are supported under WinFS. To support Win32 applications querying WinFS you will even be able to use paths for arbitrary queries. \\localhost\store\contacts\surname\williams etc.
"Taligent is still pure vapor. Maybe they'll be the last who jumps up on Openstep... "
Well yes it will, but I wouldn't park that in the filesystem as such, any more than I'd consider the locate db to be part of the filesystem.
I'm wondering whether part of the problem is that you and I just have slightly different definitions of 'filesystem'.
To me, a metadata indexing service is a service, a filesystem is an on-disk data structure that records info about the files written on the disk. I personally would not feel that the filesystem as such has to know about metadata, beyond what it already handles. As far as I know, there's nothing wrong with adding this sort of thing as a service. Just because one might wish to label it 'the central database' or 'central storage' does not mean it actually has to be an integral part of the filesystem in the sense of forming part of the description of the physical structure on disk. All that does is to increase the risk that disks are going to be unreadable by older or alternative operating systems, which, if you'll excuse my French, sucks.
That said, there is a distinct advantage in having the service catch writes/reads on the FS. Does that make the service part of the filesystem?
I'm satisfied that the MS version will/would work, though much less satisfied about its use on heterogeneous networks.
But the original poster was actually issuing a challenge along the lines of Write A WinFS-alike filesystem under Linux, and I was saying that I personally wouldn't attack the problem from that direction, but would prefer to write a metadata indexing/robust linking/history system that runs as a service. As for the paths, isn't that essentially just a URI handler connecting to the service?
But WinFS is just a service. It doesn't change the on disk structures, thats still NTFS. The metadata is cached into an SQL Server Yukon based database that is stored in a hidden folder.
They are changing the file system driver code, but only to make it more transactional to ensure the database and the file system remain consistent with one another. WinFS itself just runs as a usermode service, albeit a very heavyweight one in the early Alphas. I doubt the actual on disk layout will change at all, though they will probably start to make more use of existing NTFS features like multiple datastreams and junctions.
"Taligent is still pure vapor. Maybe they'll be the last who jumps up on Openstep... "
In other words, MS and I do not share our definitions of the term 'filesystem'.
.net) its major distinguishing feature was a requirement for more CPU and memory than my machine had... but I suppose it's 'eating one's own dog food' from MS's point of view.
Last time I ran SQL server (with WinXP and