Hacker News Reader

Comment by lotrjohn | Hacker News Reader

Reader beware. Article is from 2022 and stuff has probably changed a lot. OpenBSD is currently in version 7.9, article talks about 6.6.

Also the PC Engines APU2 the author speaks about is EOL and no longer in production.

I do like how the article is clearly dated at the top. I believe all internet articles should be dated - it provides a huge amount of context imo.

View on HN

I doubt that anything could have changed, because OpenBSD has not implemented any new file system and a few years ago they have even removed the support for "UFS with soft updates", which is a file system that is not affected by power failures, but the OpenBSD developers could not maintain it, because apparently today nobody understands it any more.

Thus there is no doubt that the UFS variant used by OpenBSD remains very sensitive to power failures, so this will remain a persistent problem, until they add a more resilient file system.

The right technical solution would have been to merge a good Linux file system, e.g. XFS, but that is impossible because of license conflicts.

The alternative is to take ZFS from FreeBSD, which might be OK, but I do not know this for sure, because on FreeBSD I use only "UFS with soft updates" which works fine for me and it is still immune to power failures.

I have some knowledge of the BSDs and they all are but OpenBSD is particularly belligerent about Licenses and they consider ZFS "non–free" on par with the GPL(I know I know) which is a huge barrier to entry into the kernel. They only begrudgingly accepted clang after they changed their license because there was no real alternative. Before that they had been using some ancient GPLv2 gcc.

Truth is, I've never had a catastrophic data loss on traditional filesystems like ext2 and ufs and the devs probably haven't had one either so they don't feel like the situation is the same as with lacking a compiler.

I read the article about why they removed soft updates and the gist of it was that the maintainer moved on and no one else was willing to understand the complex system anymore.

I'd imagine ZFS is at least on the same level of complexity.

So, two pretty tall hurdles to clear.

If someone gave them a filesystem with the right license, ideas from 20+ years ago and a promise to maintain it on OpenBSD forever, I'm sure they'd take it.

>If someone gave them a filesystem with the right license, ideas from 20+ years ago and a promise to maintain it on OpenBSD forever, I'm sure they'd take it.

They have the right to take NetBSD's UFS (with WAPBL), FreeBSD's (UFS with softupdate) or DragonFly's Hammer2.

Or give users the freedom to use ZFS (optionally) like NetBSD.

> they consider ZFS "non–free" on par with the GPL

Why isn't it? The CDDL is basically the MPL, which in-turn is basically the LGPL.

The jihad against the GPL by FreeBSD and conveniently ignoring the CDDL being weak-copyleft was always ironic to me.

> and the devs probably haven't had one either

solene@ left and stated the filesystem as a supporting reason:

> I have grievances against OpenBSD file system. Every time OpenBSD crash, and it happens very often for me when using it as a desktop, it ends with file corrupted or lost files. This is just not something I can accept.

https://dataswamp.org/~solene/2024-11-15-why-i-stopped-using...

> The jihad against the GPL by FreeBSD and conveniently ignoring the CDDL being weak-copyleft was always ironic to me.

CDDL is not viral.

Honest question: Am I the only one bothered by the fact that this author puts her name absolutely everywhere on her posts? Feels very ego-trippy.

Reply on news.ycombinator.com