DragonFlyBSD Lands New EXT2/3/4 File-System Driver

[tuxd3v]

I've worked in the computing industry for over 25 years. I see more FreeBSD in deployment today than ever before. Why? not sure, if it's replacing other Unix's like Sun or HP or it fills some technical gap, or older engineers like it. There are a lot of commercial platforms based on it too. Expensive stuff you'd never know about unless you worked in the industry.

Use to be a rarity, I remember seeing one or two FreeBSD 4 box around.. and I also worked for a company that had OpenBSD deployments for firewall/VPN.. then around FreeBSD 8-9 release I started to see a lot more..

NetFlix, uses and abuse of BSD, Facebook frontends, the same,
A lot of companies goes with BSD for firewalls/Load Balancers duo to the excellent network stack, and security they offer..

But above all, I believe BSD starts to be the good old GNU/Linux, with superb network, and security, and ofcourse the KISS aspect of it,
GNU/Linux starts to be mud waters, troublesome, a hog, and BSD is capitalizing on that, in the server space..

Because in the server space you want something practical, functional, logic, secure, and well implemented..without big burden/complex init systems above it..
The server is there to rock,
Not the waist sysadmins time, and waist processing power/resources doing ...what people don't want too..

[k1e0x]

NetFlix, uses and abuse of BSD, Facebook frontends, the same,
A lot of companies goes with BSD for firewalls/Load Balancers duo to the excellent network stack, and security they offer..

But above all, I believe BSD starts to be the good old GNU/Linux, with superb network, and security, and ofcourse the KISS aspect of it,
GNU/Linux starts to be mud waters, troublesome, a hog, and BSD is capitalizing on that, in the server space..

Because in the server space you want something practical, functional, logic, secure, and well implemented..without big burden/complex init systems above it..
The server is there to rock,
Not the waist sysadmins time, and waist processing power/resources doing ...what people don't want too..

It's not just Netflix, tho that is a well known one, and pretty impressive really. What I'm talking about is the big products you'd never hear of. (sorry if I'm vague, gotta be..) I mean like.. the kind of software where you can only sell it to a dozen customers in the world. You know.. like you have to be Walmart size to even need it. Some of that software is FreeBSD.

And I agree with a lot of what you said. Companies still need simple quick things. FreeBSD is a good fit there. So were as Unix itself isn't as popular as in the 80's and early 90's, unless you count Linux as Unix, (depends how hard you are squinting) FreeBSD itself is more popular today then ever.

[aht0]

Careful, that there are a lot of loud mouth primadonnas in charge of BSDs too.

I can only think of Theo de Raadt. Otherwise generally there's "live and let live" and/or "No interest in what they do" attitude towards Linux.

That does not speak very nicely of any of the BSDs, changing the on-disk format for a basic filesystem is regrettable.

Times and computers were much different when UFS was originally designed. At some point (could have been 2013) data structure alignment was altered (getdents system call was adapted to 64-bit ino_t). That in turn forced change in on-disk format. Pile on top NFSv4 ACL's, Soft Updates+Journal, snapshots etc which were added after Linux's UFS implementation went on "maintenance-only" mode and you should get the picture.

I agree on the general statement that there are bullshit or even barely maintained filesystems in Linux, although the good ones with some form of support are still more than those in other OSes.

But looking at "default kernel configs" is bullshit, there are A LOT of features that only make sense for some types of hardware, or some architecture or some usecases. You can't assume that all stuff that isn't on by default is barely maintained and broken.

Default kernel config gives you rough estimation about what an average user is thought to be needing. From that POV it's usable metric tho.

[pal666]

Oh it has lots to do with attitude, starting from top down (Torvalds, Hartmann.. etc), ending with, well - you. Users tend to mirror leaders of their favorite projects.

you are projecting

Linux's "supporting dozens of filesystems" is a hype.

it's a fact. it supports dozens of filesystems not invented in linux, i.e. there's no nih syndrome involved.

Huge percentage of these "supported" file systems are historic cases, experimental ports (often read-only) and so forth

that takes us back to my previous post: code doesn't write itself. if you need your fs to improve, you have to do it

. Read: lot are useless for anything except for bragging rights.

that's something that bsd idiots do

That same UFS2 support in there is literally useless, unless you happen to have 15 years old FreeBSD install somewhere you want to access. It hasn't seen any work, excepting maintenance since before last decade. Meanwhile on-disk-formats have changed in all the BSD's. But you can claim you have one more file system "supported", dont ya?

because bsd idiots complain instead of writing code. i've already given you irix xfs example.

Take a look at default kernel configs and see what's by default enabled - and that's much more limited selection. But that gives you plain truth.

$ ls /lib/modules/5.7.8-200.fc32.x86_64/kernel/fs/ -1
9p
affs
afs
befs
binfmt_misc.ko.xz
btrfs
cachefiles
ceph
cifs
coda
cramfs
dlm
ecryptfs
erofs
exfat
f2fs
fat
fscache
fuse
gfs2
hfs
hfsplus
isofs
jffs2
jfs
lockd
minix
nfs
nfs_common
nfsd
nilfs2
nls
ocfs2
orangefs
overlayfs
pstore
reiserfs
romfs
squashfs
sysv
ubifs
udf
ufs
vboxsf
xfs

FYI: there's actually current (last updated on April 26th 2020) Hammer v1 (userspace)driver for Linux (https://github.com/kusumi/lh1)
There are couple more I know of which were abandoned by authors

most common reason for startup failure is "no market need"

[pal666]

I say unfortunate because crossplatform ZFS and BTRFS in the works means less and less reasons to consider BSD

isn't it fortunate?

[pal666]

Illumos itself was at fault by being too timid accepting PR's. That, not mystical technical merits of Linux kernel were the reason of switch.

what are you smoking? linux kernel didn't accept zfs prs and will never accept them

I am curious, are you selectively de-humanizing certain people in your mind? There was guy literally half a dozen posts earlier asking about H2. I think he would count as "someone"?

it's statistical noise and that guy is passive, and as we already know, code doesn't write itself, so only "someone who writes code" counts

[pal666]

You don't need to use FAT32, ZFS is cross platform. FreeBSD, Linux, MacOS, Windows & Solaris/Illumos

as you could easily see in my list above, fat is supported by linux, zfs isn't

[pal666]

The real problem is that the two most successful desktop operating have really poor support for alternative file systems out of the box.

shouldn't that problem be alleviated by wsl2(i didn't check)?

[pal666]

Anything that runs on the a BSD runs just as well on Linux, so no point in using it unless you want native ZFS

the only os where zfs is native is solaris

[pal666]

Nope, I am simply suggesting nothing more or less than NIH syndrome combined with sense of superiority and incompetence over subject matter

lol, that's exactly how we've got all incompatible *bsd forks

Ages a go I found following from FreeBSD Forums where this very same was discussed.

you and poster just forgot to look into a mirror