Reiser4 File System Still In Development 317
An anonymous reader writes "Reiser4 still hasn't been merged into the mainline Linux kernel, but it's still being worked on by a small group of developers following Hans Reiser being convicted for murdering his wife. Reiser4 was updated in September on SourceForge to work with the Linux 3.5 kernel and has been benchmarked against EXT4, Btrfs, XFS, and ReiserFS. Reiser4 loses out in most of the Linux file-system performance tests, has much stigma due to Hans Reiser, and Btrfs is surpassing it feature-wise, so does it have any future in Linux ahead?"
Rename it (Score:5, Insightful)
If you can find any name that's not related to murdering your wife, go for it. Bonus points if it's catchy.
Time to let it go... (Score:4, Insightful)
Re:Time to let it go... (Score:5, Insightful)
What's also key is that the better points of ReiserFS, such as journaling, have migrated into other file systems. The experiment wasn't a failure, it was a darn good idea that has led to an overall improvement in reliability and speed of other file systems.
Why would you even care? (Score:5, Insightful)
has much stigma due to Hans Reiser
Really? You can't just judge it based on it's features and performance?
So if Linus Torvalds ever commits a crime, you'll stop using Linux?
Re:Time to let it go... (Score:5, Insightful)
I've been watching Btrfs and it feels like they're merging in most of the features Reiser had in mind without saying so explicitly. I've considered it a spiritual successor for a while now.
Re:my personal opinion (Score:5, Insightful)
As far as I can tell btrfs has what features would be worth taking and then some - and is under more active development. I think reiserfs is dead long-term.
I ran reiser3 for a long time. I was happy with it. But these days I'm on ext4 (and eyeing btrfs). Wonderful thing about btrfs and the ext* FS's is that they provided a migration plan. Reiser4 (at least last I checked) could not convert an existing FS (even reiser3). btrfs can even convert ext4 and allow you *TO GO BACK* if you want. How awesome is that?
Re:Why would you even care? (Score:4, Insightful)
It's human nature. His name taints his work now. Most people don't want to look at shit while they're eating nor do they want to think of some asshole who killed his wife when formatting a file system.
Re:Rename it (Score:4, Insightful)
Either it's a decent codebase or it isn't. If it is, there's no point in wasting all that labor just because one of the authors is a murderer. It's like if you wanted to ban every murderer's memoir from distribution. It's a silly approach IMHO. Don't anthropomorphize the code. Your admission "I'd rather let him watch his brainchild die" is not a rational response at all. Personally, I'd much rather exploit the fruits of his labor if at all possible. That's a bit more productive, don't you think?
Re:Time to let it go... (Score:4, Insightful)
Databases obviously handle it just fine, so the theoretical impossibility isn't there, and demonstrably so. The question is whether existing filesystem APIs are up for the job. I think they aren't. Passing things back and forth one file name or one file descriptor at a time is quite wasteful. Syscalls aren't free.