Got a tip for us?
Let us know
Become a MacRumors Supporter for $50/year with no ads, ability to filter front page stories, and private forums.
Apple Shuts Down Open Source ZFS Project
- Thread starter Small White Car
- Start date
- Sort by reaction score
I don't expect Apple to attempt to retrofit a fs running on Linux because it's easier.
The design goals are totally different here. Linux is primarily found in the biz sector and thus any fs is going to be tailored for business and not consumers or media.
Apple's focus is on consumer application and multimedia. They've had a filesystem team that works fulltime and that's not on ZFS.
What they really need to do is not necessarily rewrite from scratch what they have in HFS+ but rather to add the features that benefit their needs. Snapshots really aren't necessary if you have a solid Time Machine. Preventing "silent corruption" with checksum really isn't that necessary. What is necessary is the fs taking on some of the tasks of storing data that that today is strewn across the OS.
Who here looks at .ds store littered around and thinks "geeze this is archaic"
Apple's fs may be smarter about storage as well. It may have features that allow it to be tailored appropriately for dealing with solid state storage and spinning rust.
I think the advancement of 10.7 and on is going to depend on what a new filesystem brings.
The design goals are totally different here. Linux is primarily found in the biz sector and thus any fs is going to be tailored for business and not consumers or media.
Apple's focus is on consumer application and multimedia. They've had a filesystem team that works fulltime and that's not on ZFS.
What they really need to do is not necessarily rewrite from scratch what they have in HFS+ but rather to add the features that benefit their needs. Snapshots really aren't necessary if you have a solid Time Machine. Preventing "silent corruption" with checksum really isn't that necessary. What is necessary is the fs taking on some of the tasks of storing data that that today is strewn across the OS.
Who here looks at .ds store littered around and thinks "geeze this is archaic"
Apple's fs may be smarter about storage as well. It may have features that allow it to be tailored appropriately for dealing with solid state storage and spinning rust.
I think the advancement of 10.7 and on is going to depend on what a new filesystem brings.
It's a good thing.
I'd rather have Apple roll their own instead of trying to retrofit Sun's creation for the sake of appeasing the tech blogging elite (who don't know jack **** about databases to begin with). It's just a buzzword for them.
ZFS is good but there are legacy considerations like HFS+
You can't just switch to ZFS and have everything magically work without causing major headaches for developers. Apple engineers are good but they aren't magicians.
I'd like to defer judgement on Apple's motives and ignore the unfounded nerd rage since Apple better knows what's good for their future roadmap.
I'd rather have Apple roll their own instead of trying to retrofit Sun's creation for the sake of appeasing the tech blogging elite (who don't know jack **** about databases to begin with). It's just a buzzword for them.
ZFS is good but there are legacy considerations like HFS+
You can't just switch to ZFS and have everything magically work without causing major headaches for developers. Apple engineers are good but they aren't magicians.
I'd like to defer judgement on Apple's motives and ignore the unfounded nerd rage since Apple better knows what's good for their future roadmap.
Are you kidding me? Over half of OSX is opensource in one way or another. The filesystem would not even be on the top ten list of most controversial codebases.
Yep, but no part of the Core OS is GPL. GPL programs running on the OS are fine, but it is a definite no no to put a hunk of GPL code in the core. The filesystem most definitely would be noticed by FSF and bring a response.
No, the Core os, by which I guess you mean Darwin and BSD are mostly BSD or Apple-licensed, but that changes nothing. They still have to open up any changed code if asked. They wouldn't have to do anything different with BtrFS. Let's say they make a fork of BtrFS, like they did with KHTML/Webkit, and it works better in OSX than the Oracle codebase did. So what if they had to give that code to anyone who asked? It's not like the totally separate kernel which USES that module of GPLed code would have to be compromised.
It's all about how you program it. That's why you can make commercial software that runs on Linux, made in Java, with a MySQL backend.
Not entirely unexpected. Bummer all the same. I think Apple is making a mistake if they go to an in house system. Open source FS is the way to go. I don't necessarily care if it's ZFS, but supporting an open standard helps everyone.
From wikipedia:
The principal developer of the ext3 and ext4 file systems, Theodore Ts'o, has stated that ext4 is simply a stop-gap and that Btrfs is the way forward, having "a number of the same design ideas that reiser3/4 had"
I've always thought of Reiser as the "killer" filesystem...
From wikipedia:
The principal developer of the ext3 and ext4 file systems, Theodore Ts'o, has stated that ext4 is simply a stop-gap and that Btrfs is the way forward, having "a number of the same design ideas that reiser3/4 had"
I've always thought of Reiser as the "killer" filesystem...
Meaning what, exactly? Copland was entirely an Apple project, terminated when Steve Jobs brought NeXTSTEP with him and decided to go that route instead. Why wouldn't Steve Jobs have brought NeXTSTEP with him along with those developers when NeXT ran aground and he returned to Apple? It was certainly an in-house project, but it's not like it was a random acquisition.
Huh? It's been three years for ZFS. "Longhorn" was six years in the making.
I don't really see the point you're trying to make with either example. Whether Apple will be successful in creating a new file system has little to do with either--as with all development in the industry, they will hire the developers or buy a project already in development and go from there. Almost nothing springs up out of thin air.
That's a bit of an oversimplification. They haven't "gotten it right" because they haven't decided what "it" is. Microsoft has made some pretty significant progress in that time, but until they stop playing with conceptual features and settle on what NTFS' replacements needs to be, of course they won't ship a replacement.
Wirelessly posted (Mozilla/5.0 (iPhone; U; CPU iPhone OS 3_1_2 like Mac OS X; en-us) AppleWebKit/528.18 (KHTML, like Gecko) Version/4.0 Mobile/7D11 Safari/528.16)
So is everyone else
macduke said:
So is everyone else
Retrofit ? That's not how creating file systems works. Think of a computer file system as the skeleton/nervous system of an animal. It's not something you design after you've designed the rest of the animal.
Microsoft's WinFS relational file system was supposed to be in W95 when they started working on it 20 years ago.
It didn't even make it into W7.http://en.wikipedia.org/wiki/WinFS
For Apple to create a from the ground up new file system would take at least 10 years or more. My guess is we won't see a new file system from Apple or MS for at least another 5 years.
ZFS was the best hope for Apple to get there first and parts of it were already in Leopard. Maybe shutting dow the open source project means more internal development for OSX.
OS X is in desperate need for a new file system. I say this as last week I needed to restore my Whole PC as the inodes got corrupted and said my disk was full, and the tools couldn't fix them.
OS X really needs a Fast modern File system that can do the following...
Incremental copies and modifications, As well the same for backups. So when changing your 32gig virtual machine by say logging in it will not backup 32 gigs of data. For a few added bytes.
Self check and correct for disk errors. Laptops are apples biggest Mac items. That means a lot of bumped drives that sometimes gets errors that needs to be corrected. (No more bad iNodes)
OS X really needs a Fast modern File system that can do the following...
Incremental copies and modifications, As well the same for backups. So when changing your 32gig virtual machine by say logging in it will not backup 32 gigs of data. For a few added bytes.
Self check and correct for disk errors. Laptops are apples biggest Mac items. That means a lot of bumped drives that sometimes gets errors that needs to be corrected. (No more bad iNodes)
I would like to think that and hope you are right, but won't that take a lot longer to create a file system? HFS has been around for like 25 years or something, right? If it was that easy it seems like they would have done it already. I am afraid it will be more like 10.8 or 10.9, but I hope you are right.At first this sounds bad, but then it seems like Apple is working on their own next-generation file system.
One would presume, this being Apple, that they're going to make sure it has all the advantages of ZFS plus a few more goodies. This might take longer but I'm betting it will be worth it.
Could be the biggest feature of 10.7.
I am disappointed to see this news.
I agree.
I was disappointed to read that ZFS is dropped because I was hopeful for a new sophisticated and more robust file system for the Mac, but I agree that Apple knows the best way. I hope that they are adopting some best of breed open source project already well underway. I have no clue what that should be, Reiser, or Ext4 or whatever. I just hope for something good that is better and more reliable.
They still have to open up any changed code if asked. - nope - not for BSD.
KHTML is LGPL and can be linked with commercial software / software with an incompatible license. Your use of the word USES is called "linking" and is specifically mentioned in the GPL. Look at the Linux kernel and the arguments over device drivers and blob (which were given a exception by Linus Torvalds).
Commercial software running on top of Linux is fine. For MySQL you must purchase a commercial license (MySQL is dual licensed) to use the libraries. This is one of the problems with Oracle buying Sun, Oracle's ability to kill MySQL's commercial option by not doing licensing.
Time Machine works on the file-level, ZFS on filesystem level. This means that it can more easily and more importantly automatically track changes. Really, they are nothing alike.
WinFS is not a file system. WinFS runs on top of NTFS, which is a file system.Microsoft's WinFS relational file system was supposed to be in W95 when they started working on it 20 years ago.
WinFS was a data management system.