Apple Pulls the Plug Completely on ZFS | Hacker News Reader

Apple Pulls the Plug Completely on ZFS

zfs.macosforge.org

103 pointsdurin4247 comments

There's a Google Group [0] and a github repo [1] for those interested in keeping going with the last code drop from Apple.

[0] https://groups.google.com/group/zfs-macos

[1] http://github.com/peaceful/zfs-mac

I ported it and have an installer for Snow Leopard:

http://dustin.github.com/2009/10/23/mac-zfs.html

Five years on and ZFS still remains just out of reach. I'm not a likely user on OS X (anyone got a good use case for laptop use?) but other filesystem + volume management solutions on Linux never seem to match up.

Here's hoping btrfs keeps it real.

Don't know, Gruber mentions something about ZFS being now related to Oracle as a bad thing, and the ties of btrfs and Oracle may do this another no-go from legal

btrfs is GPL, which is an automatic no-go for Apple legal.

I'm curious, where does the Apple/Oracle bad blood speculation come from? I don't remember seeing any references to it before, so was Gruber just grasping at things? The NetApp lawsuit seems way more plausible.

Much of the lower levels of Mac OS X (including the compiler, which at the moment is gcc, although Apple is moving toward LLVM/clang...) is GPL. CUPS, which is now managed and primarily maintained by Apple, is GPL.

Realistically, disk filesystems have to be linked into the kernel (FUSE isn't going to cut it: too slow, and can't be used on boot), and Apple isn't going to link GPL code into their kernel. All the stuff you listed is userspace.

i'd say "much" is exaggerated. More like "in a few places where they couldn't rip off MIT/BSD licensed code". It's a shame that Apples success i based on a lot open source code but they give so little back to the community.

Give so little back? Really?

libdispatch was something purely in-house. They're giving it away. Same for launchd, although other groups inexplicably feel a need to recreate it from scratch. They're employing major contributors to LLVM/clang and having them work fulltime on those projects. CalendarServer is a fantastic CalDav server, which they're giving away.

Don't say they give "so little" back to the community. While I may really hate what's going on with ZFS right now, your statement wasn't really fair to them.

Probably harsh, but nevertheless my point of view. When i look at http://www.wolframalpha.com/input/?i=microsoft+vs.+apple+vs+... and compare apples size in terms of employees, revenue... it's a _huge_ company with lots of money. Now i look at what it does in the open source world... From my perspective a lot of what i see on macs today has it's origins in the open source world and apple wouldn't be where it is today with open source (webkit came from khtml, mac os x kernel partly from freebsd, calendar server is based on twisted, afaik). Now you tell me that the company that tries to lock it's customers in their ecosystem with a lot of proprietary software&hardware (hello quicktime, hello itunes), that it's particularly open source friendly? It sure as hell is not. I could nearly bet that MS does more open source then apple. If it's not already the case, apple is heading in that direction.

Probably harsh, but nevertheless my point of view.

Your view seems to be informed by ignorance. For instance, your claim that they've only given back where they couldn't "rip off" BSD/MIT-licensed code is pretty specious when Apple's largest contributions are to BSD-licensed projects that they use. Who exactly is getting ripped off?

it's a _huge_ company with lots of money.

What differences does that make? Your prior criticism was that they took much from open source and gave back little. When you bring money into it, you're making it sound like their contributions should be based on what they have, not what they have taken. That doesn't seem very fair given how many of their employees and how much of their revenue has very little at all to do with things taken from open source.

A more reasonable perspective is that they should give back in proportion to what they took, rather than that they should give back in proportion to how much money they have. The former is fairness, the latter is entitlement.

webkit came from khtml

That was seven years ago. Longer than KHTML even existed before Apple's fork. A lot has changed in that time. Today, much of the development on KTHML is backported from WebKit, which is itself a vastly more active and widespread open source project than KHTML ever was. Half a dozen projects and millions of users have benefited from Apple's development efforts. I don't see how this can be interpreted as anything less than an open source success story.

(hello quicktime, hello itunes)

I'm not sure what you're getting at. Both of those products use standard non-proprietary formats by default.

I could nearly bet that MS does more open source then apple. If it's not already the case, apple is heading in that direction.

Could you cite examples of Microsoft's open source contributions for comparison? Or explain how Apple is "heading in that direction"?

If Apple wants it then, btrfs' license is irrelevant to Apple as long as there isn't patent encumbrance. Apple can dispatch a team of flying monkeys to clone a btrfs implementation under whatever license they choose.

wc linuc-2.6.31/fs/btrfs/*.[hc] says 50k lines of text. Call it 20 flying monkey years of coding.

Me too. Ted Ts'o (ext4's maintainer) has said the btrfs is the future.

Just to make sure: is the guy happily married? ;-)

I guess this was a reference to Hans Reiser killing his wife. Do you really think domestic violence is a laughing matter?

Everything's a laughing matter if it happens to you, so by the Golden Rule it should be a laughing matter if it happens to someone else.

How about checksums to detect corruption on those fragile spinning disks you're picking up and carrying around the room? Or quick and efficient snapshots, scheduled frequently? This is seriously disappointing, if not surprising.

Another useful laptop feature is the option to use duplicate blocks as a poor man's RAID (zfs set copies option). Could be used for system files or other critical data.

"anyone got a good use case for laptop use?"

Their laptops are gorgeus.

OK... The OS is not my favorite. It's currently a tie with OpenSolaris for me (nicer GUI on one, package management on the other), but both bot are behind Jaunty (and probably will be behind Karmic)

I believe OP was speaking about the filesystem specifically.

Very good sleep, hibernate, and wakeup behavior.

I would not call it out of reach. You can download OpenSolaris and use it. The Gnome desktop is a bit older than Ubuntu or Fedora, but, still, it's usable.

anyone got a good use case for laptop use?

I may be in the minority but I would have liked the opportunity to use the same filesystem on my OSX laptop that I do on my database server.

Well, that's fairly disappointing. I've been excitedly watching for it to make it into the operating system. I think it would be a positive but subtle shake-up for the industry.

Damn, was looking forward to ZFS. Are they working on an alternative?

According to Gruber, yes.

http://daringfireball.net/linked/2009/10/23/zfs

Here's your chance in building the better alternative. ;-)

http://jobs.apple.com/index.ajs?method=mExternal.showJob&...

Unfortunately, if Apple dropped ZFS because of the NetApp lawsuit that means their homegrown next-gen filesystem will probably avoid many of the efficient-yet-patented techniques found in ZFS.

They could license the patents. NetApp has a 9.6B market cap, so it would not be an easy takeover target.

Except that Apple has $30B+ in cash. Still, Apple could roll their own for much less than the $9.6B NetApp is worth.

How much could they roll their own for, few hundred thousand? Depends on timescale and starting point I suppose. But let's call it $1Million USD. That surely puts an upper bound on licensing the tech needed from NetApp to continue with ZFS. Perhaps they just think ZFS sucks?

Yeah, right. And they have been hanging to HFS+ for all this time because they knew it's a solid base for future development...

Sorry for the sarcasm, but I couldn't resist, but the very idea of Apple making a storage system like ZFS out of thin air is very... optimistic.

I wouldn't expect them to use ZFS. It would be a really messed up court case when NetApps licensed the patents so Apple could use ZFS which NetApps was suing Sun / Oracle for a patent violation (head hurts). I would expect they would do a license to cover their own implementation of a file system.

$1Million for a next-gen filesystem? The programmer wages for the testing phase will cost 10 times that.

Hans Reiser made a couple for no wages ... I guess that muddies the waters when considering the baseline for me.

Writing a file system isn't the hard part. Testing and verifying that it actually works under all kinds of strange scenarios and to make really really sure it won't corrupt your data, no matter what. That's the hard (and expensive) part. I seem to recall reiserFS being plagued with a number of bugs that only showed in uncommon corner cases that trivial testing wouldn't find.

why?

This is Apple we're talking about. The world may never know.

Looks like it was a legal issue:

http://daringfireball.net/linked/2009/10/23/zfs

That's too bad if so. ZFS is pretty cool technology (i've been pretty impressed with it so far for my personal use). I was really hoping it was going to be added at some point to OSX. Hopefully apple's in house implementation will fix a lot of the same issues ZFS was aiming at.

Ah yes, patents - the enemy of progress.

I'm all for hardware patents (they seem to get licensed and used), but these software ones have to go.

The Firewire patent(s) arguably killed Firewire, in an analogous way to software patents.

I think you mean the license fees killed Firewire. Apple got greedy and lost.

It's the same thing. The purpose of the Firewire patents was to collect licensing fees.

Seems like this decision came from Oracle. Switch to freebsd. =)

Noooooooooooo.

Reply on news.ycombinator.com