Conversation

Notices

1. Embed this notice
   Baloo Uriza (baloouriza@social.tulsa.ok.us)'s status on Friday, 01-Sep-2023 10:45:26 JST Baloo Uriza

   Well, that was certainly a brash byline that lacks context. ReiserFS was second only to ext3 when Reiser went off the deep end 20 years ago. ext4 almost didn't happen and certainly delayed ext3's maintenance and further development, because of ReiserFS's growing popularity at the time.

   @kevinpurdy

   https://arstechnica.com/gadgets/2023/08/the-torrid-saga-of-reiserfs-nears-its-end-with-obsolete-label-in-linux-kernel/

   • Doughnut Lollipop 【記録係】:blobfoxgooglymlem: repeated this.
   • Embed this notice
     feld (feld@bikeshed.party)'s status on Friday, 01-Sep-2023 11:03:36 JST feld

     in reply to
     Reiserfs and reiser4 were victims of cancel culture. Both were technical marvels
     Doughnut Lollipop 【記録係】:blobfoxgooglymlem: likes this.
   • Embed this notice
     feld (feld@bikeshed.party)'s status on Friday, 01-Sep-2023 11:08:26 JST feld

     in reply to

     • Johnny Peligro
     Reiserfs has an amazing tail packing feature that made it scale better than anything else when it came to a ton of small files.
     
     Reiser4 was basically on the road to being the best parts of XFS/ext4/btrfs IMO.
     
     Their research papers they'd release on their design work were great too
   • Embed this notice
     Johnny Peligro (mischievoustomato@marsey.moe)'s status on Friday, 01-Sep-2023 11:08:27 JST Johnny Peligro

     in reply to

     • feld
     i am too lazy to read wikipedia. why were they technical marvels?
   • Embed this notice
     feld (feld@bikeshed.party)'s status on Friday, 01-Sep-2023 23:34:51 JST feld

     in reply to
     Hans Reiser AND Edward Shishkin were NameSys.
     
     Edward was basically blackballed by the community after Hans murdered his wife.
   • Embed this notice
     Baloo Uriza (baloouriza@social.tulsa.ok.us)'s status on Friday, 01-Sep-2023 23:34:53 JST Baloo Uriza

     in reply to

     • feld

     @feld LOL, no dude, cancel culture had nothing whatsoever to do with it. Hans Reiser *was* NameSys. Without him, there was no company. Without the company, there was nobody working on ReiserFS.

   • Embed this notice
     feld (feld@bikeshed.party)'s status on Friday, 01-Sep-2023 23:37:21 JST feld

     in reply to

     • feld
     Reiser5 was announced by Edward in late 2019. But you wouldn't know that because everyone intentionally ignores ReiserFS stuff now like it's radioactive.
     
     https://www.phoronix.com/news/Reiser5-Development
   • Embed this notice
     Sexy Moon (moon@shitposter.club)'s status on Friday, 01-Sep-2023 23:56:21 JST Sexy Moon

     in reply to

     • feld
     @feld @BalooUriza probably should have changed the name to something less bad like EpsteinFS
   • Embed this notice
     feld (feld@bikeshed.party)'s status on Saturday, 02-Sep-2023 00:13:28 JST feld

     in reply to
     Filesystems take at least a decade before they can be considered stable, so I'd say it's very relevant vs stuff like bcachefs which I wouldn't touch for a long long time
   • Embed this notice
     Baloo Uriza (baloouriza@social.tulsa.ok.us)'s status on Saturday, 02-Sep-2023 00:13:30 JST Baloo Uriza

     in reply to

     • feld

     @feld This far past the half life of Hans arrest, I'm guessing it's far less radioactive as irrelevant.

   • Embed this notice
     feld (feld@bikeshed.party)'s status on Saturday, 02-Sep-2023 00:14:43 JST feld

     in reply to
     The business didn't need to continue for the filesystem to continue.
     
     And Edward did continue -- he was hired by RedHat to maintain reiserfs/reiser4 from 2009-2014
     
     Now he's paid by IBM to continue filesystem/storage research.
   • Embed this notice
     Baloo Uriza (baloouriza@social.tulsa.ok.us)'s status on Saturday, 02-Sep-2023 00:14:44 JST Baloo Uriza

     in reply to

     • feld

     @feld This seems to conveniently ignore that Edward had no access to accounts and effectively no say at NameSys, to such a high degree that NameSys offices never reopened after Hans arrest.

   • Embed this notice
     feld (feld@bikeshed.party)'s status on Saturday, 02-Sep-2023 00:15:08 JST feld

     in reply to
     ext4 is terrible lol
   • Embed this notice
     Baloo Uriza (baloouriza@social.tulsa.ok.us)'s status on Saturday, 02-Sep-2023 00:15:10 JST Baloo Uriza

     in reply to

     • feld

     @feld Whoosh. I'm saying ReiserFS missed its opportunity, ext3/4 already ate its lunch.

   • Embed this notice
     feld (feld@bikeshed.party)'s status on Saturday, 02-Sep-2023 00:28:53 JST feld

     in reply to

     • Phantasm
     Its multithreaded performance is still ass compared to XFS, which solved its small file problems a long time ago so it basically does everything you need
     
     fs shrinking is insane, nobody should use that. If you're stuck in a position where you *need* to shrink an fs you did something very very wrong and you should be punished for not planning better or owning more hardware to alleviate your issue
     
     BTRFS is also CoW and it's not super useful to compare CoW filesystem so log/journal filesystems as they have very different use cases. e.g., putting a database on a CoW is really dumb and you'll just suffer the write amplification penalty
   • Embed this notice
     Phantasm (phnt@fluffytail.org)'s status on Saturday, 02-Sep-2023 00:28:55 JST Phantasm

     in reply to

     • feld

     @feld @BalooUriza I wouldn’t say it’s terrible, at least not now. I’ve had really good success with it for basic stuff and hardware caused fs corruption recovery. It’s basically the NTFS of Linux filesystems. It gets the job done, but lacks interesting features from others (deduplication, ssd optimizations, snapshots), but has some that the alternatives don’t.

     For example:

     • XFS is faster, but cannot be repaired online, AFAIK there’s no fs shrinking
     • BTRFS corrupts itself if you run out of space and use compression, RAID 5/6 is broken for many years now, slows down to a crawl if you use quotas and big snapshots
   • Embed this notice
     feld (feld@bikeshed.party)'s status on Saturday, 02-Sep-2023 02:59:27 JST feld

     in reply to

     • Phantasm
     Shrinking is just an incredibly complex and dangerous feature to do, so it's not surprising most filesystems do not support it.
   • Embed this notice
     Phantasm (phnt@fluffytail.org)'s status on Saturday, 02-Sep-2023 02:59:29 JST Phantasm

     in reply to

     • feld
     @feld @BalooUriza
     
     >fs shrinking is insane, nobody should use that.
     It has it's uses. It's insane to handle on the FS and partition level, but should be supported at least while offline. Shrinking on a server/workstation is insane, that I agree. For home usage where you are more likely to need it I think it's reasonable, if you have backups.
     
     For example you can't predict how much space you would need on a laptop for a Windows partition 5 years down the line, if you dual boot. Worst case, you can wipe everything and restore from backups for 6 hours as you extract everything from external media.
     
     >BTRFS is also CoW and it's not super useful to compare CoW filesystem so log/journal filesystems as they have very different use cases.
     Agreed. I put it there only as an example of features comparison and it's problems.