Chimera Linux Pushes Ahead For FreeBSD User-Space Atop Linux, Built Using LLVM

[justinkb]

It's good this exists. Their use of another toolchain will inevitably lead to bugs in the software on offer being discovered. I myself have found quite a few bugs in widely used software when I was using LLVM toolchain to build my system on exherbo. More often than you'd expect developers will rely on implementation details of gcc/libstdc++ instead of what standard specifications actually say.

[AusMatt]

Doesn't FreeBSD already use the LLVM toolchain to compile it's userland (base system and packages)? If yes, then the value is just about compiling the kernel with the LLVM toolchain. Aren't there already projects that do this?

I'm concerned that this project may encourage bugs to reside in its (inevitably forked) FreeBSD userland. Maybe they plan on upstreaming their changes to require running with a Linux kernel... but are FreeBSD devs interested in maintaining such support?

[sinepgib]

FWIW, the main point of of the meritocracy line in the presentation was to expose every single asshole that I want to stay away from the project early on, and it seems it has been pretty successful

I have enough work to do as it is, and I'm not about to get demotivated by additionally having to deal with toxic people

Are you the author of the project? I had a question that went unnoticed under all the noise, I wanted to know if I'm missing something technical or if it's a just for fun project. What motivated you to start this project?

[q66_]

Are you the author of the project? I had a question that went unnoticed under all the noise, I wanted to know if I'm missing something technical or if it's a just for fun project. What motivated you to start this project?

i'm already on the core team of another distro (void) and over the years i've accumulated all sorts of annoyances (most of which are present in most distros) that are at this point unfixable without starting over from scratch, so i mostly just wanted to create something fresh that does not suffer from 30 years of technical debt, is simple and straightforward, but does not take on either the "veteran unix sysadmin" or "vague minimalism" smell

i don't really want to imitate anything else, and most technical choices in the system have some specific reasoning behind them (e.g. using clang unlocks hardening possibilities which would've been impossible with gcc as well as makes cross-compiling extremely trivial + makes the system LTOable without introducing insane compile times and memory requirements, a from-scratch buildsystem means a secure, introspectable packaging collection that is also extremely fast to build, BSD core utilities mean high quality code that is reasonably featureful unlike the likes of busybox, and so on)

[aht0]

lol. freebsd userland is where you can't put -l after ls *.txt

your point being..?

ahto@SERVER ~> ls *.txt -l
ls: -l: No such file or directory
1a.txt 22.txt 23.txt 24.txt 2a.txt
ahto@SERVER ~ [1]> ls -l *.txt
-rw-r--r-- 1 ahto wheel 0 Feb 16 00:01 1a.txt
-rw-r--r-- 1 ahto wheel 0 Feb 16 00:02 22.txt
-rw-r--r-- 1 ahto wheel 0 Feb 16 00:02 23.txt
-rw-r--r-- 1 ahto wheel 0 Feb 16 00:02 24.txt
-rw-r--r-- 1 ahto wheel 0 Feb 16 00:02 2a.txt
ahto@SERVER ~>

[pal666]

your point being..?

ahto@SERVER ~> ls *.txt -l
ls: -l: No such file or directory
1a.txt 22.txt 23.txt 24.txt 2a.txt
ahto@SERVER ~ [1]> ls -l *.txt
-rw-r--r-- 1 ahto wheel 0 Feb 16 00:01 1a.txt
-rw-r--r-- 1 ahto wheel 0 Feb 16 00:02 22.txt
-rw-r--r-- 1 ahto wheel 0 Feb 16 00:02 23.txt
-rw-r--r-- 1 ahto wheel 0 Feb 16 00:02 24.txt
-rw-r--r-- 1 ahto wheel 0 Feb 16 00:02 2a.txt
ahto@SERVER ~>

my point being it works perfectly on sane operating systems with any order of arguments

[aht0]

my point being it works perfectly on sane operating systems with any order of arguments

Doubt that -l would work in front of command.. argument busted. Definition of sanity is also defined by surrounding environment. Nutcases are pretty "normal and sane" in insane asylum.

[pal666]

Doubt that -l would work in front of command..

i said any order of arguments, not arguments before command

Definition of sanity is also defined by surrounding environment

for command line it's sane to accept arguments in flexible order because sometimes people decide to add new argument after they typed previous ones and simplest solution is to add it without looking for correct place to insert. linux is user-friendly, freebsd is not, because it doesn't have enough manpower to implement even such simple thing as convenient argument parsing

[aht0]

i said any order of arguments, not arguments before command
for command line it's sane to accept arguments in flexible order because sometimes people decide to add new argument after they typed previous ones and simplest solution is to add it without looking for correct place to insert. linux is user-friendly, freebsd is not, because it doesn't have enough manpower to implement even such simple thing as convenient argument parsing

Pseudo-issue to bitch about while Linux has weirder problems.. like idiotic way of designating device drivers. I've never had urge to type arguments at the end of an expression but.. while looking at network driver designation I got urge to swear. Can't easily tell which fucking network device is which by just looking at them. eth0/1/2 was manageable but not perfect, newer way is brain dead - maybe just assign them SHA512 hashes to make idiocy perfect?

Basically, I am playing around with OpenSUSE Leap 15.3 right now. Had to manually rename network device designations to be BSD-style (for example i218 nic is now igb0 and older pro1000 quad nic is em0-em4) to be able to always tell with a glance which is which. Linux designations, which might have contained MAC address for all I know, was just utterly nothing-telling nor easily rememberable.

And while bootup takes like 5 seconds, why the fuck shutdown still takes 30s+?

While we are at it.. explain me this shit

# cat /sys/fs/ext4/features/casefold
supported
# mkfs -t ext4 -O casefold -E encoding_flags=strict /dev/nvme0n1p2
mke2fs 1.43.8 (1-Jan-2018)
/dev/nvme0n1p2 contains a ext4 file system
last mounted on Sat Feb 19 23:35:07 2022
Proceed anyway? (y,N) y
Invalid filesystem option set: casefold
#

So, is the damned casefold supported or not? My assumption is that, while casefold feature is in OpenSUSE kernel, mkfs doesn't know about it, being from 2018 itself. This kind of inconsistent shit I keep running into, is one of the reasons I seriously dislike Linuces.

[pal666]

Pseudo-issue

lol, that's how you call lack of basic usability

Linux has weirder problems.. like idiotic way of designating device drivers.

linux has correct way of designating device drivers, but uneducated freebsd zealots can't understand it

Can't easily tell which fucking network device is which by just looking at them. eth0/1/2 was manageable but not perfect, newer way is brain dead - maybe just assign them SHA512 hashes to make idiocy perfect?

well that's the only way you can tell which is which, other ways like eth0 don't tell you anything at all(and can get reordered on reboot)

Linux designations, which might have contained MAC address for all I know, was just utterly nothing-telling nor easily rememberable.

lol so you still don't understand what those designations mean. don't worry, in linux you can rename or alias interfaces

And while bootup takes like 5 seconds, why the fuck shutdown still takes 30s+?

it's your system, you should know it. shutdown gracefully shuts down all services. my system shuts down faster than boots up, on your system something takes long time, only you can investigate it

While we are at it.. explain me this shit

# cat /sys/fs/ext4/features/casefold
supported
# mkfs -t ext4 -O casefold -E encoding_flags=strict /dev/nvme0n1p2
mke2fs 1.43.8 (1-Jan-2018)
/dev/nvme0n1p2 contains a ext4 file system
last mounted on Sat Feb 19 23:35:07 2022
Proceed anyway? (y,N) y
Invalid filesystem option set: casefold
#

So, is the damned casefold supported or not? My assumption is that, while casefold feature is in OpenSUSE kernel, mkfs doesn't know about it, being from 2018 itself. This kind of inconsistent shit I keep running into, is one of the reasons I seriously dislike Linuces.

is ext4 with casefold supported on freebsd?
i don't use suse, on fedora it works

$ mkfs -t ext4 -O casefold -E encoding_flags=strict file
15:28:31
mke2fs 1.46.3 (27-Jul-2021)
Discarding device blocks: done
Creating filesystem with 102400 1k blocks and 25688 inodes
Filesystem UUID: 4bf37efd-8ca5-4c0c-90d7-95918da7b075
Superblock backups stored on blocks:
8193, 24577, 40961, 57345, 73729

Allocating group tables: done
Writing inode tables: done
Creating journal (4096 blocks): done
Writing superblocks and filesystem accounting information: done

so maybe your main problem is bad distro choice?