What is Digg?70 Comments
- Rudegar, on 08/01/2008, -12/+40wife's of developers beware :O :P
- RandyFithian, on 08/01/2008, -1/+24This is the great thing about open source, new and interesting alternatives coming out just for the sake of innovation.
- wolferz, on 08/01/2008, -0/+18No he was making a very bad pun about the the murder of Reiser's wife... by Reiser.
...and is some how getting dugg for it. - Breepee, on 08/01/2008, -1/+18Please, read up on ZFS. Read up on it's checksumming, constant integrity checking, automatic versioning (which are usefull for everyone, including homeusers), not to mention ease of use when doing more advanced stuff like raid and multi-users environments.
ext3 is Linux' NTFS: a table of where all the files are with journaling. Sure, it's reliable and does what it does well, but for a 21st century filesystem it does too little. Let's move all things filesystem into the actual filesystem, and let it worry about backups, integrity and all those things. - Giga, on 08/01/2008, -0/+14Surely you aren't suggesting we give up and stagnate just because it works well enough, are you? What's wrong with letting someone else do the hard work to try to find a better alternative, and if they fail, so what? We learn something.
- Loonacy, on 08/01/2008, -1/+14You really shouldn't refer to ReiserFS as "killer".
- Protoss, on 08/01/2008, -4/+15Could have picked a better name. This sounds too much like a distro or something, not a filesystem...Maybe I'm just used to my filesystems being some seemingly random string of letters and numbers.
- neFariou5, on 08/01/2008, -4/+14Just a shame Windows would never submit to a unix filesystem. I really wish MS would do WinFS, NTFS is really holding back everything at the moment.
- neko, on 08/01/2008, -1/+10Okay.
I hope this ends up being a manslaughter filesystem like ReiserFS, before it was abandoned. - SteveMax, on 08/01/2008, -0/+8The GPL doesn't allow ZFS to be implemented, due to the patents over it. As it is now, you can implement ZFS in Linux in userland, using tools such as Fuse; implementing it on the kernel would break the kernel's license.
- wolferz, on 08/01/2008, -7/+15... what?
Last I checked the current implementations of NTFS are some of the best file systems out there. As for WinFS, NTFS at a rudimentary level already supports everything that WinFS does... WinFS just implements indexing and metadata extensions to special parts of the file system (that already exist under NTFS) rather than keeping it all in databases like Vista does now. - wolferz, on 08/01/2008, -5/+13First off... you have no evidence of this because, assuming you have a 100 gig hard drive, that text file would be taking up 65 gigs which in and of itself is ridiculous.
Second, while fragmentation does happen quite a lot under NTFS, it's not as significant an issue as it would be under other file systems. NTFS does an excellent job of indexing the locations of data in such a way as to be almost completely unaffected by fragmentation. The only limitations then are hardware and seeing as modern hard drives spin at 7200 rpm OR MORE this is a non-issue as well except in cases where large volumes of data are read and written to the disk constantly over prolonged periods of time (large database and web servers). For home computers and workstations fragmentation is a complete non-issue. - wolferz, on 08/01/2008, -0/+8I see no reason not to try to improve, even forking or starting again from scratch.
However the parent comment seems to suggest that the lack of an alternative is some how a bad thing rather than just an average thing. It's the difference between need and want. I want a better filesystem than the ext*fs'... but I don't need them. - RedCt, on 08/01/2008, -3/+9If it works out to be a good ext* alternative, great. It's not like we have too many...
- theaceoffire, on 08/01/2008, -0/+5"One file system should be enough for anybody!"
- aladrin, on 08/01/2008, -0/+5Right, because 'Tux3' says exactly what it is... Nobody would ever mistake it for anything else.
At least if it was cryptic, people wouldn't assume it's something that it's not... They'd just wonder what it is. - mynameistux, on 08/01/2008, -1/+6you fail
- basickler, on 08/01/2008, -0/+5I'd love any implementation of ZFS for linux. I built a 48 drive RAID on solaris using it and it made mdadm+lvm look like brain surgery. 5 commands, and a couple minutes of my time and I had a filesystem that was redundant, had snapshots and built in compression. Not to mention that I've seen how long it takes to rebuild broken raids with 20 some odd disks in Linux. On ZFS this is a non issue, combined with the built in checksumming it makes previous implementations of software raid look pretty sad. Unfortunately this is countered with solaris being quite a pain to use (IMHO) and buggy as hell on Sun's x86_64 hardware (at least the 2 different boxes I've used). I'd be willing to use ZFS with fuse but it seems development kind of stopped...
Now, and this might be getting greedy, but what I really want to see is this. Sun integrates Lustre perfectly with ZFS then releases both under the GPL. Networked raid anybody :) - dotorg, on 08/01/2008, -3/+8I hope this ends up being a killer filesystem like ReiserFS, before it was abandoned.
- akatsuki, on 08/01/2008, -1/+6Ignoring all the other stuff that went on, why not just take up Reiser4 since it is basically dead in the water right now? Why reinvent the wheel in this case?
- NTolerance, on 08/01/2008, -0/+5Agreed. The feature set of NTFS is quite amazing considering its age. A lot of its features still aren't being utilized by Windows. Ars Technica has a great article about this:
http://arstechnica.com/articles/paedia/past-presen ... - ArchAngel21x, on 08/01/2008, -1/+5And the reason they feel the need to create a new file system instead of adopting ZFS is?
- wolferz, on 08/01/2008, -1/+5cryptic naming conventions are not a good thing... even if they are the norm.
- rnawky, on 08/01/2008, -7/+11Oh really? Because that new text file I just created got fragmented across 65% of the HDD.
- jgtg32a, on 08/01/2008, -4/+8Actually the Windows "Kernel" is damn good its just all the other stuff they put on it is the problem
- Darkhacker, on 08/01/2008, -0/+3All file systems are OS independent. Even NTFS can be used on Linux with NTFS-3g on FUSE. Nobody wants to use HFS because it pretty much sucks (even Apple fans will admit that). Ext3 is open source and under the GPL. Plus BSD has FFS (aka UFS) which is about as liberal as you can get with a license before going public domain.
Trying to create a "FAT64" or "OpenFAT" would be a horrible idea. FAT is not a very good file system and the only reason it's used on portable devices is because it's easy to implement and supported on most major operating systems. There is nothing to stop Microsoft or Apple from implementing an open source file system though and there is nothing to stop device makers from using it. - unrealmp3, on 08/01/2008, -0/+3*hint at ReiserFS*
- mrBitch, on 08/01/2008, -0/+3developers certainly have many wives - we pick them from our harem of groupies...
- SteveMax, on 08/01/2008, -0/+3Changing the kernel's license would be basically impossible. You would need the permission of every person who has ever contributed to it, and many are unreachable or even dead. The only possible license change is to a newer version of the GPL (still not easy to do, since there is code that is licensed as "GPLv2 only" and not "GPLv2 or later"), but GPLv3 only complicates the matter further.
Adopting ZFS in Linux, as it is, is really impossible. *BSD kernels are BSD licensed, and don't suffer from any of the GPL limitations; this means they could implement ZFS without any legal question.
The GPL is the problem in this case, but going away from the GPL is not possible. You can't even fork GPL code to some other license unless you have the permission of all authors (so making a CDDL or BSD-licensed Linux fork is impossible).
This is a case where the GPL really blocks every possibility. Only if Sun re-licenses ZFS under the GPL, but even then they'd possibly go for GPLv3 (so the kernel would have to be re-licensed to be compatible with it). - Elranzer, on 08/02/2008, -0/+2It's too bad that Microsoft will never implement a non-Microsoft filesystem, or Apple would ever add write-support for NTFS in OS X. It's great that Linux can read/write pretty much every filesystem, but they're the only ones.
We're stuck with FAT32 on portable devices due to Microsoft/Apple's stubbornness. - SteveMax, on 08/01/2008, -0/+2Technically, you still need authorization to change their code, since they keep the copyrights. We'd have to contact their family/family lawyers to get a signed authorization.
Really, it's not going to happen. - OCPaul, on 08/01/2008, -0/+2I though it was a game loL Tux3. Anyways hope to try it one day.
- KristofferR, on 08/01/2008, -1/+3The ReiserFS murderer
- FolkTheory, on 08/01/2008, -0/+2Comrade!
- SniperZero, on 08/02/2008, -0/+1The graphics were good but not good enough to kill someone.
- brnews, on 08/01/2008, -1/+2Shouldn't it be called TuxFS?
- bnolsen, on 08/02/2008, -0/+1In my experience reiserfs was much better for handling huge file systems. Running fsck on an ext3 array that big is unacceptable and goes more than a business day. reiserfs was always faster for checking & fixing and in my experience much more robust to recovery from intermittent hardware faulting.
Sorry I've only dealt with ~600 disks in raid arrays. - basickler, on 08/01/2008, -1/+2Umm, EXT3 is like the bread and butter of linux file systems and I don't see that changing any time soon. But ZFS has a lot of great features that make it entirely different. Just to mention a few, checksumming (no real need for raid battery), built in volume management (no more LVM), snapshots (back to LVM), redundancy (no more mdadm), and built in compression just to mention a few.
Try managing arrays of 30 or 40 disks in a multi-user environment and you'll quickly see ext's many limitations. One of the most pertinent being the 16TB partition limitation. - dazparkour, on 08/01/2008, -0/+1They want it to be snap shottable and to save Versions of files - this isn't for the sake of innovation, read the article - it's for the sake of keeping up with the other file systems that ALREADY do these things.
I do however LIKE the idea of having both of these - although, I use Subversion anyway. - neFariou5, on 08/05/2008, -0/+1Indexing within a file system is a big deal, it basically allows for a tagging-like system when saving files, meaning you no longer have to save everything in sub folders to keep organised. You'll quickly find anything on your computer, and you don't need some Indexer Service running all the time in order to do it.
- theaceoffire, on 08/01/2008, -0/+1Cause average joe knows what "NTFS" means.
- AnImAl6969, on 08/03/2008, -0/+1Half the fun of this article was reading the comments.. Much more interesting then the story itself..
- bradkovach, on 08/01/2008, -0/+1That's right, wolferz: have you ever even built a 16 TB partition? basickler just handed your ass to you.
- gemmakicn, on 08/01/2008, -0/+1I suspect that it would take someone of reiser's mental fragmentation to continue his work...
- Atomic1fire, on 08/01/2008, -0/+1Its the 3 P's
Penguins Practicing capital Punishment - mrBitch, on 08/01/2008, -0/+1Speaking of how powerful ZFS is, it's amazing to me how quietly Apple has added ZFS to the feature list for their upcoming Snow Leopard update to OS X.
- wolferz, on 08/02/2008, -0/+1@theaceoffire
and I repeat... EVEN IF THEY ARE THE NORM.
Do you speak English? - Atomic1fire, on 08/01/2008, -0/+1Rieser did, well until the divorce... and the murder
- geniusj, on 08/01/2008, -1/+1HammerFS was just released for DragonflyBSD which basically accomplished the same goals.
- theaceoffire, on 08/01/2008, -1/+1Author of Reiser4 has bad PR right now.
Better to start fresh I think. -
Show 51 - 70 of 70 discussions
The Digg Toolbar for Firefox lets you Digg, submit content, and keep track of Digg even when you're not on the Digg site. Download the official 
© Digg Inc. 2009
— Content created and posted by Digg users is