Linux Developers Discuss Deprecating & Removing ReiserFS

[Raka555]

For original ReiserFS I have to agree no benchmarking is necessary, it's simply obsolete, however people report to be still using it. Let's see if anybody comes to the mailing list during the deprecation stage.

The filesystem might be stagnant but changes to the virtual memory subsystem,block layer and other parts of the kernel can have a substantial influence.

[Etherman]

I installed a system on reiserfs root yesterday! 😀 A VM with Funtoo, for fun. I didn't notice any problems.

Hope someone makes a reiserfs-fuse-thing before it disappears from kernel so old filesystems can be accessed.

[mdedetrich]

The original ReiserFS was quite revolutionary, it treated the filesystem as a database which gave it quite novel capabilities.

However due to the obvious unfortunate history and the amount of time that has progressed, its fallen behind and there are arguably better designs for more modern filesystems.

[Volta]

Ext4 was not the fastest either. In fact the early ext3/ext4 filesystem were extremely slow.
20 years of optimizing might have paid off.

I remember old Phoronix benchmarks and ext4 was usually the fastest.

[wooptoo]

The original ReiserFS was quite revolutionary, it treated the filesystem as a database which gave it quite novel capabilities.

However due to the obvious unfortunate history and the amount of time that has progressed, its fallen behind and there are arguably better designs for more modern filesystems.

The original ReiserFS was ahead of its time and definitely faster on my Pentium III machine than ext3 at the time.

Thanks for posting the gnome2 screenshot Michael, brings back some good memories

edit:

Benchmark: https://linuxgazette.net/102/piszcz.html
You have to take into account the hardware at the time rather than compare it to today's ext4.

[King InuYasha]

Edward is an active maintainer, so it's unlikely to be deprecated.

Edward maintains the out of tree Reiser filesystem, not the in-tree one. The in-tree ReiserFS code has no maintainer. You can see in the MAINTAINERS file in the kernel sources yourself:

REISERFS FILE SYSTEM
L: [email protected]
S: Supported
F: fs/reiserfs/

[skeevy420]

I last tested reiser4 on rotating media and at that time it wiped the floor with all the other filesystems.

Only thing that prevented me from using reiser4 was that it was not in the mainline kernel.
I had my share of non-bootups after a kernel update.

But with SSDs the emphasis moved from disk layout to parallelism .

Ext2 is not an option these days because of the lack of journaling, but ext2 had insane parallelism back in the day.
(Also not sure if it is even still in the kernel)
It might actually be an option on laptop since Apple also play with fire: https://mobile.twitter.com/marcan42/...13855387734019

Back in 2016 I was debating between Reiser4 and BTRFS for my long-term data drives since both offered inline compression. Due to having negative experiences with BTRFS I was leaning towards Reiser4. Right when I was about to pull the trigger is when I discovered OpenZFS. It offered everything I thought I needed and more. I'm going on my 6th year as a ZFS user now.

Since we're talking nostalgia and back in the day in this thread too: How many of y'all were running Windows 2000, fired up a Longhorn beta, and thought to yourself, "If that's the future I'm jumping ship" and OS hopped until you found Linux?

[Sin2x]

Edward maintains the out of tree Reiser filesystem, not the in-tree one. The in-tree ReiserFS code has no maintainer. You can see in the MAINTAINERS file in the kernel sources yourself:

REISERFS FILE SYSTEM
L: [email protected]
S: Supported
F: fs/reiserfs/

I'm not sure how this works officially but Edward does in fact maintain the in-tree ReiserFS and has posted a patch to fix the issue a day after it was posted on the mailing list. Maybe he will also add support for Y2038 later as well, as kernel continues to ignore Reiser4/5.

[King InuYasha]

I'm not sure how this works officially but Edward does in fact maintain the in-tree ReiserFS and has posted a patch to fix the issue a day after it was posted on the mailing list. Maybe he will also add support for Y2038 later as well, as kernel continues to ignore Reiser4/5.

He hasn't submitted Reiser4/5 upstream at all, so I'm not sure how that's the kernel "ignoring" it.

[Sin2x]

He hasn't submitted Reiser4/5 upstream at all, so I'm not sure how that's the kernel "ignoring" it.

Really? https://reiser4.wiki.kernel.org/inde...4_and_upstream