“Imagine a Windows operating system that makes the physical location of your files irrelevant. Imagine that, regardless of where you save a file, you can search for it, find it, and open it from any location on your computer–so that anytime you need to organize, access, or otherwise manage any file on your system, it’s all just one viewing pane away. That’s the concept we see in early builds of Windows Longhorn, the OS Microsoft says will merge all flavors of Windows XP, and, more importantly, overhaul the Windows file system for the first time since Windows NT.” Read the article at News.com.
An Early Look at Windows Longhorn’s New FileSystem
About The Author
Eugenia Loli
Ex-programmer, ex-editor in chief at OSNews.com, now a visual artist/filmmaker.
Follow me on Twitter @EugeniaLoli
It is good t see Microsoft inovating again.
Are they saying they created this concept yet?
-Bruno
can’t we do that already. Start / Search / Find (Windows XP already runs an index on everyfile to speed up document lookups)
It just seems as though they are trying to dance with the fact that they are basically incoprating base level SQL Server functionality into their new file system.
This is just another reason why now SQL Server will outperform Oracle and the others (on windows that it) …
It’s just a database taking the place of your filesystem. This sort of thing has been going on for a very long time in the business world.
I believe this is fairly worthless for even the average user. Who here disables Windows Indexing services under Windows 2000/XP?
hehe perky little brazilian ;D
I’m quite sure this is the extend part of the “embrace & extend” tactic ;D
it will make organising and moving files into folders much faster since all it will require is a link being moved.
so basicly your files will now just be a logical not a physical arangment and having links to the files you should see no loss of performance since the link will just be a file that stores the file’s db index. I think it will be a good thing but it could end up being a bad thing if they impliment it poorly.
I every day use an operating system with a database on-top of the filesystem.
I use it from more than three years and to short I will also acquire one new version of my preferred operating system.
I wanted also to answer Bruno, the first likeable commentator, for “thanks” him for the work that he carried out, that is the realization of the opensource version of the BFS.
For a long time I use its replacement and of they are much satisfying, I also find it better of the original BFS.
You excuse to me if I do not write nothing of that I imagine, but as many others I so’ what mean and for to be really useful a such innovation it also needs of the rewriting of the existing applications.
Unfortunately Microsoft introduces it why they to Palladium works, but this will not be a my problem!
Imagine that, regardless of where you save a file, you can search for it, find it, and open it from any location on your computer–
Well I got to get it now, hehe. They are partly doing the new file system to keep liux from reading windows.
Presumably this will be a case of Windows 95 Syndrome.. where a new product turns out to be an old product with a new skin…
My query, assuming this – and the posting above from deb-man – is whether real file access will be possible for essential tasks such as file backup. I guess it will have to be possible in some format since interoperability would be impossible without……
They are partly doing the new file system to keep [linux] from reading windows.
Bingo!
s/partly/mostly
“They are partly doing the new file system to keep [linux] from reading windows.”
Not worth the time and effort. Hopefully WinFS is faster than NTFS. Browsing and refreshing certain directories takes too long sometimes.
When I use Windows I never have a problem finding
anything.
If you can’t orgnaize your files, a good find feature will
take care of if for you.
Maybe this will prevent the windows fs need for defragging.
As Miss Daisy would say. “Ridiculous”
This file system can be very nice, but it will take some re-learning to make it useful. If you don’t use it the way it is meant to be used, it will only make your system slower.
this sound like an enterprise version of Lotus Notes.. didnt they say that the lotus notes idea was a really bad one ?
My most important files are text based – essays, programs, ebooks and I very carefully organise them in folders and make multiple backups. If people were to carefully organise their data, wether documents, MP3s, pictures, then I don’t see the need for this particularly when there’s probably a CPU and memory overhead.
The file system is trying to make up for poor organization. Solve the latter and you don’t need the first.
only because it used the NFS. They’ve now moved to a DB2 back end after openly admitting that their suite sucked. Thank God I say.. I’m tired of constantly restoring the damn mail files.
This is a good thing for dev’s though. I really don’t think this is a huge improvement for the typical joe blow user. But programmers (depending on how you hook into it) will probably find it much nicer to work with. However, now we’re fiddling with the fact that programs/programmers will still have to support “legacy” file systems = bloat.
Whether this is any advance over a simple ‘search by file extension’ will depend on how much information it can extract from the files on your drive, as the average user is not going to bother creating metadata for each file.
Closed file standards will cause MS big problems here, while it may be able to catalog using the embedded info in mp3s, AVIs and MSWord files, will it be able to extract title and format information from a Quicktime film?
Or would the media or format it cannot recognise just not appear when you search, except as a filename with a generic ‘movie’ catagorisation and size in KB?
If, as the article says, “its indexing system runs continuously in the background.” I really hope this can be turned off, as it will make the system unuseable for audio work.
” its indexing system runs continuously in the background. “
yes just imagine
the incredible clusterf**k Microsoft will make of this idea
the massive lockin they will try to achieve
the huge amount of downtime that will result when your “database” filesystem corrupts
ahhh yes…just imagine!
http://m-arriaga.net/software/newdocms/
It’s already been done for KDE. Nice to see MS catching up again 🙂
it will make organising and moving files into folders much faster since all it will require is a link being moved.
Just like on an inode-based filesystem. Whyt don’t they just go for a totally POSIX filesystem then?
Or is this another way to slow the competition down by not letting access to MS drives?
It will be interesting to compare WinFS with ReiserFS and the work that some are doing on the ext3 with ACL’s in the not so distant future. Just so everyone is aware … Hans Reiser has been working on these ideas for more than a decade in Unix/Linux filesystems. He is also being sponsored by DARPA (you know, the agency Al Gore started to invent the internet
http://www.namesys.com/whitepaper.html
This whitepaper was written in 1984 and revised in 1993. Gives you an idea on how much MS is really “innovating” in this regard
Windows Storage Full-Text Index
I wonder how full this is going to be. Will you be able to search for “wha’ the f**k?” for example? (including quotes and punctuation).
Indexing doesn’t work on my W2K because it seems to want to index brand new file instantaneously. So, when I create a temp directory during a build, odds are I won’t be able immediately delete the directory, as the indexer will have latched on to one of the irrelevant files and won’t let it be deleted.
Meanwhile, while I was trying to use indexing, I was quite excited to learn that it pretty much completely ignored the vast amount of data in my Outlook Express Inboxes (which is what I’d REALLY like indexed). This made it even more worthless.
Now, I occasionally run updatedb in cygwin to get a snapshot of the data on my drives, then I can use located to feed things into grep if I need too.
While rich filesystems can be a nice thing, Mac users have for years dealt with the difficulties of transferring and maintaining the meta data maintained outside of the files themselves.
Of course, there is zero motivation right now to come up with a standard way of bundling, tagging, and organizing rich file meta data for transport.
Finally, it will be interesting to see how well MS pulls of the moving and linking paradigm that inituitively prevents users from make N actual copies of data, rather than N actual links. When folks normally move files, they either move them completely, or duplicate them. We’ve had the shortcuts for a while, but I think people still find them confusing to actually use, and not all application actually support them.
Isn’t this idea similar to the Plan 9/Inferno concept?
The whole “hierarchical filesystem” is getting antiquated. For example, I was recently trying to decide a directory hierarchy for my music files. Should I organize them {Artist/Album/Title} or just {Album/Title}. Then I realized, heh, what do I care, and just dumped them into a big directory and let my jukebox software (JuK for KDE) take care of giving me multiple views into the organization of the filesystem. This power is why programs like iTunes and picture browsers are popular. Now, imagine extending this power to the entire filesystem. There are no longer any directories. What appear to be directories are just a list of files based on a query. So my /home directory becomes a powerful window into all my documents, instead of just a static structure containing /home/code, /home/docs, /home/media, etc. Now, I hope Microsoft implements this the right way. SQL seems a little too meaty for something that works at the filesystem level. Also, I hope they don’t pull a “OS X” and expose the database interface only at the GUI (Explorer) level, rather than in the lowest levels of the filesystem. I wanna be able to do queries from my CLI
I don’t know much about Reiser4 (the next-gen version of ReiserFS, due out this summer), beyond what’s in the (detailed) whitepaper, but it appears they have a plugin architecture that includes support for arbitrary directory semantics. If that is the case, Reiser4 might be very well suited for implementing something like this.
Somehow this all sounds very familiar to me,as it would any BeOS user,WHy this could be even bigger than “The Media OS” WinXP(where I have personally witnessed that with a 1.5Gig CPU and 128 megs ram you still can’t run more than one audio app at a time without the sound skipping and sticking
These Boys sure are the cutting edge LOL (at stealing other people’s technology and making it screw up that is!)
One question – what is Windows Indexing Service used for ????
You guys are living in a time warp. Err the reinvention of the Amiga concept. Bah!
1. This is not a new idea.
2. It is a “solution” looking for a problem.
3. Creating a system that creates a logical abstraction of a physical implementation is a perfectly good idea. Basing such a system on SQL Server is a bad idea because of problems with:
(a) the physical implementation (its crash prone);
(b) security (every upgrade seems to open a new huge security hole and reverse the security patches applied to the prior version); and
(c) logical interface (SQL Server, without application specific code, cannot guarantee accuracy as to all operations on the data).
As to a prior comment — [Also, I hope they don’t pull a “OS X” and expose the database interface only at the GUI (Explorer) level, rather than in the lowest levels of the filesystem. I wanna be able to do queries from my CLI
] — Mac OS X has the terminal for CLI access and the Finder for folder or tree metaphors.
almost everybody said “This whitepaper was written in 1984 and revised in 1993. Gives you an idea on how much MS is really “innovating” in this regard
”
Come on, who cares who innovates what ? Big companies almost NEVER innovate on technical issued, as it comes always from university. Did apple invented the GUI ? No. Did Sun invented Unix ? No. Was object oriented languages invented bu firms ? No. Simula67, and smalltalk were university projects at the beginning. But apple spread GUI, ATT spread object oriented concpets with C++, etc…
Microsoft made a pc for everyone possible, but as microsoft is today evil ( some computer science teachers of mine even compare Bil Gates to Hitler… No comment! ), everybody blame it for eveything.
someone also said a lot pf crap like
“Somehow this all sounds very familiar to me,as it would any BeOS user,WHy this could be even bigger than “The Media OS” WinXP(where I have personally witnessed that with a 1.5Gig CPU and 128 megs ram you still can’t run more than one audio app at a time without the sound skipping and sticking
”
Well, under win2000 with my PIII800, I can do a lot of musical stuff. I can play Virtual synthetiser whith almost no latency ( a few ms ), record to 100 audio tracks, [email protected] khz, etc… You just don’t know what you’re talking about. A lot of musicicians do professionnal stuff with PC, today.
beOS was very promising, but today, is there just ONE person who do profesionnal music on it ? No.
Where I work we have a 300G MS-SQL database that recieves around 2M transaction per day. The rate of growth on the system is too fast because the Indexs aren’t kept clean; it seems that to keep the speed up, they didn’t bother with keeping the index clean; thus, we have to do a weekly reindex to recover disk space. MS-SQL is the only database I know other then MS-Access that needs to be rebuild to keep it working.
Now just imagine… I download a copy of GNU’s GCC source code and compile it. I then delete everything because I don’t want it any more. Guess what? I have less disk space now then I did before because the indexes are hoging the disk space. Also, just moving files around on the system will cause the indexes to grow. I just hope the include an utility to rebuild the index. So now we need the defragmentor and a re-indexer; thanks MSFT.
The idea is good but the execution has a lot to desire. 80% of the files on a hard drive don’t need to be index for easy searching. A better option would be to create a set of API’s to allow the users and the programs to register documents they want indexed. If I download a MP3, then I want to register it. If I create a word document, then I want to register it. If I download the source to GCC and compile it, then I don’t want to register it.
If you use WinFS, then the best thing to do is:
1) Create a NTFS partition to house the OS and programs.
2) Create a WinFS partition to house user data.
3) You can mount the WinFS partition under the c:Documents and Settings directory so user data will automatically be index. (I use W2K, other versions put user data in other locations).
> Imagine a Windows operating system that makes the physical location of your files irrelevant.
Oh yeah ! I can easily imagine that !
any file will be stored in M$ server which I need my Passport account to access and only from windoze…
Thanks, but NO thanks !
query ‘((name==”**”)&&(BEOS:TYPE==”audio/x-mpeg”)&&(Audio:Bitrate==”192 kbit”))’
OK, ok, maybe I’m wrong, but WHY they said that physical location doesn’t matter? In the video field, you have dedicated A/V (audio&video) drives, or better yet, a local fibrechannel SAN.
So, your system will have drives for OS, and others for video and data… now, where you stored your files? Ah… indexing services help on that.
but a SQL-like file system? um…
I will need to check this before further commentaries…
My .2 cents.
Luis
But it falls to complete crap when you consider files without much metadata. Text files, for example… What metadata do I have to query? Authors, mod/create dates, size, owners, a few other values… How does this help me find the 3rd letter I wrote to Bill related to the design project executed for the Xyz client in the first quarter of 2001? Do I want to construct such queries? No. However, it’s pretty easy to look in a client directory with quarter/year directories with a correspondance directory in it…
Just consider your completely flat structure–the only useful way to ever find something is through a query… whereas a hierarchical system would allow you to query and browse via a known filing system (whatever the user chooses and whatever they feel is best for organizing files).
I think there’s huge potential, but suggesting the hierarchical system should be abandoned entirely just doesn’t make sense to me.
[gabou ]
“Big companies almost NEVER innovate on technical issued, as it comes always from university. Did apple invented the GUI ? No. ”
XEROX Parc wasn’t a university, but a corporate research center.
[Rayiner Hashem ]
“The whole “hierarchical filesystem” is getting antiquated.”
Not so much antiquated, as overwelmed. Think information overload. Besides if we got rid of “antiquated”, were would we put the old folks?
“What appear to be directories are just a list of files based on a query. ”
Sounds like Evolution’s vFolders.
[jefro]
“Or is this another way to slow the competition down by not letting access to MS drives?”
Now you know why I have a shared partition between Windows and Linux.
[Sandy Dunlop]
“Just like on an inode-based filesystem. Whyt don’t they just go for a totally POSIX filesystem then?”
Someone once pointed out how more and more Unix-like Microsoft’s OS is becoming.
Will the db contain the files themselves (like AS/400 (?)) or just the metadata ?
I really don’t like the idea of all my data locked into a db because that would make it more difficult to move it around.
What Microsoft has planned goes a LOT further than anything out today, including BFS.
This is not necessary on Linux because all the files are stored in the same location for each user in the /home/username directory. Also organizing by hypertext is a much more useful way of organizing anyway. These people who can’t find files on their computers are not going to be helped by some magical solution. Organization takes effort and time and is a skill of its own. It would be funny to see everybody’s database get corrupted except that some of my veterans records are in Microsoft systems – I didn’t get a say on that.
I mean: anytime Microsoft tries do do anything to enhance the OS the bashers come crawling out of their holes and start yelling… When I first started reading osnews it felt like there where many that knew what they where talking about but that feeling is gone… Now it seems most of the comments are posted by crybabies (just look at the comments on almost everything Eugenia posts) 🙁
[end rant]
No, modifying the filesystem to work like this is not new, not even for Microsoft. _But_ previous attempts have been either slow and complex or having limited functionality, if MS succeeds in making this usefull for everyday tasks it is indeed innovative.
Some more comments:
. Xerox was not the first in making a GUI, they innovated many things commonly associated with GUIs but being first? Nope.
. Inode based filesystems where not first made for UNIX, is not UNIX specific and doesn’t make Windows UNIX-like.
. Everything MS does is not for f*cking with other operating systems users. (Incredible I know but sometimes they want their users to get a better system)
It’s entirely possible (and feasible) for an implementation to do text indexing, so you stuff like your “third letter to Bill” example will work. Beyond that, most people tend to think about files based on type, date, or content, which are easy to auto-index. If you’re going to go to the trouble of making a directory hierarchy with all your letters, you can go to the trouble of tagging your letters with metadata. And under a database style scheme, it’s entirely possible to emulate a hierarchical system by assigning a “parent” attribute, which could be automatically done by the system if you so choose.
As for vFolders, I think they support my point. You’ve got email software, music software, and photo software, all emulating the same functionality. The situation is basically begging to be generalized and put into the system.
As for OS X, my you supported my point. OS X looks different depending on whether you use the FileManager or the CLI. For people who just use the FileManager, it doesn’t matter, but for people who use both (me) it’s very irritating.
All you people are theorizing on this WinFS thing and nobody even knows if it will be used. It is a new file system microsoft is experimenting with and nobody but the programmers have actually tested it. Why are you jacka$$es flaming about it when you’ve not even tested it. The flaming crowd here (probably pimple faced linux nerds with no life) just have to find something to flame about with no base. I have yet to see any resonable discussion about it the filesystem, all I see are flames.
It’s too easy to back on a convicted monopolist who has an earned reputation of constantly scheming to destroy its competition.
Anyways, I suspect this release is just a bunch of trash marketing hype. Very likely it’s MS moving more of its 1970’s vintage technology up to 1990’s levels and putting out tons of buzzwords and marketing fluff for doing something more incrementally correct.
Someone said
“XEROX Parc wasn’t a university, but a corporate research center. ”
XEROX Parc is an exception : they almost never sell anything. Did Xerox earned money with GUI, ethernet, smalltalk ? I doubt that.
Every apple zealot says that microsoft just copy apple, but they are too blind to see that without xerox parc, Steve Jobs wouldn’t found apple. Apple didn’t invented the mouse neither. And really, who cares ? Without apple, it is obvious that GUI for personnal computers wouldn’t be the same today.
But I see here a lot of profesionnal people, who know everything on something they have never used, or even seen. It is pathetic.
“It is good t see Microsoft inovating again.
Are they saying they created this concept yet?”
Actually they did’nt created this concept. It’s a quite old concept and scientists already wrote about such filesystems many years ago. MS just implements it and makes it available for the desktop.
“XEROX Parc is an exception : they almost never sell anything. Did Xerox earned money with GUI, ethernet, smalltalk ? I doubt that. ”
Directly? No. However they did on the laser printer (another PARC invention).
http://www.parc.xerox.com/company/history/
I recommend people read “Dealers of Lightning: Xerox PARC and the Dawn of the Computer Age” for more details (including the whole Apple/Xerox thing).
“OK, ok, maybe I’m wrong, but WHY they said that physical location doesn’t matter? In the video field, you have dedicated A/V (audio&video) drives, or better yet, a local fibrechannel SAN.
So, your system will have drives for OS, and others for video and data… now, where you stored your files? Ah… indexing services help on that.”
Likewise for audio/music work. You don’t want 500Meg audio files mixed up with hundreds of little email or text files. If you bought a fast SCSI drive for audio mixing, it should hold only audio files.
Also, this “physical location doesn’t matter” approach goes against the principle that all storage media should be regarded as removable and hot-swappable (like CDs or Zip disks). Network connections are also temporary, especially as more portable computers with wireless connection come into use.
It sounds like a Unix-style solution to me – that, assuming that the hardware is a fixed, installed and never changed network over which the user has no control.
Maybe they should first fix the search-options of XP. It doesn’t find all files you want. Make a new file in notpad, call it test.pas, type in a phrase, e.g. ‘text’. Then search the directory for files containing ‘text’. It won’t find it! Really!
someone said
“Directly? No. However they did on the laser printer (another PARC invention).
”
Well, I said Xerox parc is an exception, because they spent a lot of money in very new tech ( like I said : ethernet, GUI, smalltalk, one of the first object oriented languages, after simula67 ), without having earned pmoney of it.
In the 70, xerox had a LOT of money, and created the parc in palo alto, without really using what was invented here. The first parc workstation was a big flop, etc…
Generally, in a R&D developpement, there is more D than R. In xerox parc, it was the contrary. No firm can live with such a sytem now, I think
By the way, my point is that you cannot blame microsoft from making better products ( only zealots can say OS like win2000 are not good, technically speaking; I personnally don’t like apple very much, I do not say everywhere OS X sucks, because it is simply not the true ), because it is not new. OF course, it isn’t, but there are now 5 relatively current OS : Mac OS / Unix / Linux / Windows / QNX. No one implements that type of FS natively ( I think it is possible to make good things with debian and HURD kernel about new concpets about files, the way they are seen by programs and userd, but it is still no very developped. I personnally think the way the files are used in linux is really a mess, and dumbed ). For example, WinNT had journaling system ( NTFS ) a lot before mac OS or linux. Has anyone said : apple are only copying microsoft by implementing journaling system in mac OS X ? It has no sense, because what does make journaling FS useful is large implementation on real like OS.
I’ve followed most of this discussion and I think something else that comes into play for a filesystem like this is personal taste. Personally I like the strict organization that a heirarchical filesystem presents. I rarely have problems finding where I stashed a file. There are other folks who find that organizing files into a strict heirarchy is limiting and would rather be able to find a file by saying, “I want the letter I wrote to So-and-so in June of last year”.
I wonder if we wouldn’t all be happy with some type of hybrid or dual-mode filesystem. Something that stores files internally in a way that is transparent to the user. The user can then choose how they want to interact with the system (classic heirarchical or new-fangled query based).
I also think that the newer filesystems that are becoming popular on Linux and other Unix platforms (such as XFS, ReiserFS, and even ext[2|3] are moving in that direction in a way with the addition of metadata. (Doesn’t BeFS support queries on metadata?)
My $0.02
A database-like file system being hailed as innovative? BeOS was doing this ten years ago!
The last thig I want on any pc, in work or at home, is MORE semi-inteligent code. Make it REALLY bright, that’s a laugh, nobody’s half managed that yet, or leave it stupid and get the user to do the work. This is just an excuse of lazyness.
Question
do you use media player, with it’s really irritating attempts to guess what I want to do, and doing it badly, or someting simple (winamp) and create your own lists and organisation etc. Now imagine the whole thing on a os level
Perhaps this is a bit paranoid, but the WinFS seems like step one for MS to control all of your data. When the new filesystem comes out, which fileservers will be able to use it? “I’m sorry but all of your Linux/Unix, Netware, etc. servers are no longer compatible, but we have this nice SQL Server setup that will run so well that your users won’t even know that their files are on the network.” Then comes offers of service: “We are setting up huge networks of SQL Server boxes so that you can store and transfer all of your data. No need to run your own servers we do all of the hard work for you and only charge you a nominal fee.” Then comes the blow at the end: “We’ve been snooping around on your computer (because our license agreement allows us to do that) and found that you are 30 seconds late in renewing your license for a single copy of Word on an unused computer on your network somewhere. Therefore we have cut off your access to all of your data. Have a nice day.”
Maybe I’m assuming the worst, but hey, it IS microsoft after all.
This article is more likely a marketing doc than a technical paper…really disappointed =<
“Imagine a Windows operating system that makes the physical location of your files irrelevant. Imagine that, regardless of where you save a file, you can search for it, find it, and open it from any location on your computer–so that anytime you need to organize, access, or otherwise manage any file on your system, it’s all just one viewing pane away.”
imagine a mac. what a joke.
Al Gore started to invent the internet
He never sad that.
stupid beos zealots don’t get it, the big part of this filesystem is the network transparency features, not just dbfs, this *IS* true innovation from microsoft.
beos is dead. long live windows.
I am surprised none of the SQL server vendors are objecting to this obvious flanking move by Microsoft.
One can imagine that given an additional hundreds of millions of seats of SQL Server that it will put IBM and Oracle at a serious disadvantage.
All the apps for Windows clients will be written to Microsoft SQL Server, giving Microsoft a power lever to move further into the datacenter.
As many can see, there are many ways to add searching and indexing to a file system. However, there are fewer ways of figuring out how to ship millions of copies of SQL Server underneath the radar…
Typical M$.
Steal what others have done (this is an almost identical copy of the original BFS which Be discovered as not being worth the hassle) and say it is *NEW*.
M$ also has one major hurdle: Backwards comptibility. Or some would say. Nope. No problems actually. They just adjust the backend of the API to handle files as if they were in folders.
In BeOS / Zeta you can see this technology in action. Moving a file is instant on the same partition. Every file (actually) is pointed to by a Node. You can think of the Node as a link (shortcut) to the actual location of the file’s data. It will actually be a great thing for Longhorn, after a few years. A worthy investment by M$, and a smart move to better their products. However, I have a strong feeling they are just going to take the oBFS code and change it a little (or at least use it as a guide). The problem is they are parading it as if it were absolutely a new thing. Predictable.
One thing they will not have however, is a developer community that knows how to effectively use the new features. Well, except the BeOS coders and other coders use to similar file systems. I do like one thing though, Microsoft is doing the system before the file system. I like this because it will cause Longhorn to not take advantage of the file system for a while. Sure, many advantages will be seen right away regardless of their coding it backwards, but many advantages of these file sytems will not be realized..ever..by anyone..
So, what can one expect with this WinFS?
(The following assumes M$ does a good job)
1. Very quick file searches
2. Ability to open the same file many times over (provided all programs have been rewritten to control node locking)
3. Ability to delete a file (node actually) while sending it over a network and that one connection be allowed to finish for security ( I wrote a security node monitor in Zeta that allows me to share files via a muscle client, and as soon as the first download to the user I wanted to get it has begun, the node is deleted, that connection continues, but no one else can ever find the file or the data (even with the original node data, as it is marked as gone except for the open transfer))
4. Ability to consildate massive programs with tons of essential files into a single clean, fast executable (attributes contain the data).
5. This file system will be much slower for file transfers.
What!?!? You ask. Yes, I say (and know).
Complex file systems such as oBFS and the future WinFS are going to be slower than conventional file systems for copying. There will also be bugs that are impossible to track (well, for WinFS anyway).
Now, one very important point to make is that the transfer speed is very dependent on the level of threading.
oBFS is very-well multi-threaded, which means you can copy many many files at once giving equal performance to each and allowing the system to perform smoothly.
WinFS may not be so heavily threaded, meaning single transfers will be somewhat faster, but try and copy a file from CD-ROM 1, to Partion 3 at the same time you are trying to copy another file from the same place to a different place, and you will very likely be pushing the limits of the file system. M$ seems to favor this so they can say: Look at the speeed!!! While only showing one transfer (maybe two if they do a decent job for M$)
Anyway, that is my forty-seven pesos.
–The loon
——————
Computer Tech, Soft Eng, Artisit (crappy one too), and a poet ( a great one if you like very dark poetry)
And a bunch of other stuff, too..
As it seems right now, WinFS will be a plain database (yes SQL Server Yukon), no metadata database grafted on a hierarchical FS. File data is going to be stored into BLOBs.
My pet hate is marketing. Why? Cause with a good marketing plan competition is NOT good for business (not bad either, just mostly irrelevant). In fact with marketing the intention is to make people sheep (follow the latest idea).
MS have a good share of the market, they have an OS that is reasonable and it works (i won’t define this word in here), so why would any user in the market purchase/invest in anything else? How many people, apart from those interested in IT, actually care who invented what technology? All that majority of people know about is what is marketed to them (unless the person has some interest in the field).
I am sure some users would do research but if they come to the conclusion that:
* they need a way to share their data with the majority of desktop market (most other people),
* lowest price
* ease of use
who will provide them with that?
Anyway, to use a metaphor, most OS debates are like trying to convince people to eat some reasonably healthy & good + convenient food (subjective opinion here but i do eat healthier and tastier food that can be served quickly) yet people still go to McDonalds/Burger King/KFC. Who has good marketing? Who doesn’t?
My opinion.