Conversation

Notices

1. Embed this notice
   Eva Winterschön (winterschon@mastodon.bsd.cafe)'s status on Friday, 23-May-2025 12:02:16 JST Eva Winterschön

   🤦♀️ Kernel Ops or Kernel Oops 🤦♀️

   - Which group makes more sense?
   - Which group tab-completes to show you all of the associated commands within the same sub-class of purpose?
   - Which group is backwards and tragic and requires rote memorization over associative structs?

   The following example is one of the many reasons why engineers prefer FreeBSD over Linux. It's the simple usability on a day to day basis.

   Cognitive load matters, and efficiency of memory matters, effeciency of keystrokes matters, and it's present everywhere. Backwards thinking is wasteful.

   kldstat
   kldconfig
   kldload
   kldunload

   ~ or ~

   lsmod
   modprobe
   insmod
   rmmod

   #freebsd #linux #kernel #engineering

   • Blaise Pabón - controlpl4n3 repeated this.
   • Embed this notice
     Eva Winterschön (winterschon@mastodon.bsd.cafe)'s status on Thursday, 29-May-2025 21:29:45 JST Eva Winterschön

     in reply to

     • Ryan Hamel

     @mrhamel I'm sure this is a joke, but otherwise, yes.

     perhaps most people are unaware just how prevalent FreeBSD is throughout the world, and part of the reason for silent unawareness is that the BSD license doesn't require the same chattiness of the GPL. so BSD often exists under the radar, hanging out on the internet backbone, posted up in critical IXPs, serving all those meaty-binary-DRM blobs from streaming services, computing the cycles of the Playstation devices, running embedded firmware the world over... or peering down and back at us from space probes...

     yes, FreeBSD kmods are actively developed, despite all of the attention that linux demands, despite all of the anger that microsoft rages on about, and still it persists at being more efficient, more stable, and more user friendly without making a fuss about it -

     FreeBSD exists without the constant interjections of the no-one-cares classic statement, "btw, I use <linux distro>" or other comedic tragedies...

     It didn't used to be this way, linux used to be great, it was inspirational, it was enjoyable, it had not yet been invaded by the worst closed-source fear-mongers the world has ever seen... but now we see these systemD/microsoft infected linux systems staggering around demanding reboots and crashing their kmods in a flaming wreckage of repeated death spirals with a garbled single last gasping demand to "LOOK AT MEEEEEEEEE!" 😆

   • Embed this notice
     Ryan Hamel (mrhamel@calckey.club)'s status on Thursday, 29-May-2025 21:29:46 JST Ryan Hamel

     in reply to

     @winterschon@mastodon.bsd.cafe Wait, engineers still make kernel modules for FreeBSD?

     Blaise Pabón - controlpl4n3 repeated this.
   • Embed this notice
     Eva Winterschön (winterschon@mastodon.bsd.cafe)'s status on Thursday, 29-May-2025 21:32:28 JST Eva Winterschön

     in reply to

     • Ryan Hamel

     @mrhamel It's funny that you think I tinker with my production systems, or with any system in the enterprise pipeline. the negging attitude isn't necessary.

     All of my corporate and enterprise work is handled though the equivalent of ITIL change control, with rigorous stages of testing which occurs from dev unit testing, through several layers of automated load-testing (hundreds of simulated workloads, batch suites, traffic replay, tens to millions of simulated client (using DPDK and TRex, among several other loadgens) with network traffic running via several Tbit/s of concurrency. Then it's off to staging, where every generation of hardware which is actively in production gets loaded and must pass multi-day to multi-week reliability validations. I don't discuss that on social media because most of it is under strict NDA.

     The majority of the load test architectures that I've built involve hundreds of machines, thousands to tens of thousands of cores, nodes costing $8K to $180K according to role and spec, with substantial budgets for a sufficient team of engineers to run the environment, while coordinating dev and production teams, all so that the org can pass compliance requirements. These architectures have absolutely influenced global infrastructure, and some of them really have had FreeBSD involved, so I'm not sure what your point is about hating on one operating system in particular, but it's unnecessary and I honestly don't care.

     The complaints that I usually discuss on social media are on test bench systems, ones that do not have global internet traffic, and most of them are mine - where I usually use rolling releases or test/edge repos. I expect breakage, but I don't care for the low standards of engineering laziness which is present in much of the modern changes to OS and service management which have occurred over the past decade.

     systemD is another story, and it's always been garbage wherever it runs. that's not debatable, just look at the CVE list.

   • Embed this notice
     Ryan Hamel (mrhamel@calckey.club)'s status on Thursday, 29-May-2025 21:32:29 JST Ryan Hamel

     in reply to

     @winterschon@mastodon.bsd.cafe Was it a joke? 😉
     
     Gone are the days of the BSD's functioning as actual routers in the Internet backbone (unless DPDK + VPP has helped to give it a comeback), and have their second chance in life as support infrastructure as you just mentioned.
     
     The difference between you and I is, I use my systems, not tinker with them. This is why I do not see the issues that you do, or do not impact me to that same degree. Also, my systems do not "demand" reboots, nor does systemd impact me as a consumer and user of an OS.
     
     Just so we're on the same page, I've met Theo De Raadt at the NANOG conference in Toronto last October, and have made several donations to the OpenBSD project for the secondary/portable software.
     
     The last sentences of your last paragraph got me thinking https://youtu.be/Lo2qQmj0_h4 + https://youtu.be/l482T0yNkeo.

     Blaise Pabón - controlpl4n3 repeated this.
   • Embed this notice
     David Chisnall (*Now with 50% more sarcasm!*) (david_chisnall@infosec.exchange)'s status on Thursday, 29-May-2025 21:32:47 JST David Chisnall (*Now with 50% more sarcasm!*)

     in reply to

     @winterschon One of the things that makes me believe the systemd people are not serious is that they decided to turn the noun-verb interface of service into a verb-noun interface in systemctl, in spite of decades of HCI research telling them to do the opposite.

     Blaise Pabón - controlpl4n3 repeated this.
   • Embed this notice
     Eva Winterschön (winterschon@mastodon.bsd.cafe)'s status on Thursday, 29-May-2025 21:33:48 JST Eva Winterschön

     in reply to

     • David Chisnall (*Now with 50% more sarcasm!*)

     @david_chisnall agreed, and it's infuriating. I wrote a translator script many years ago,, when sysD was officially in RHEL, to handle most of the generic commands that systemctl requires, simply to avoid the nonsense. I work with the init system of servers hundreds of times per day during dev/eng/test.. so those little "death by a thousand cuts" become apparent very quickly.

     the same applies to so many package managers, I ended up writing a translator so that I could reduce context switching when moving from FreeBSD to Debian/clones to RH/clones.

     I'll post the package manager and service unification translator one of these days, it's probably the type of thing which would save a lot of people a lot of otherwise wasted/inefficient brain cycles.

     the comedy of sysD never ends, and if one looks for it, the regular "service" command and regular "rc.local" and init.d stack still persists despite all of their attempts to own the system from the ground up M$FT style. I wrote a couple of ttyS getty service handlers back in.. a few months ago.. easier and faster and more reliabkle than the sysD unit file interpreted garbage.

     Blaise Pabón - controlpl4n3 repeated this.
   • Embed this notice
     David Chisnall (*Now with 50% more sarcasm!*) (david_chisnall@infosec.exchange)'s status on Thursday, 29-May-2025 21:34:38 JST David Chisnall (*Now with 50% more sarcasm!*)

     in reply to

     @winterschon

     the same applies to so many package managers, I

     The one that still always bites me on Debian-based systems is that apt install and apt upgrade do not imply apt update. So they will try doing the thing with a stale snapshot of the state of the remote repository. Often this means it will install the five packages that haven't been updated and then notice that the sixth doesn't exist (it was replaced with a newer version) and fail in the middle of modifying my system.

     Conversely, this is two of my favourite features of FreeBSD's pkg. First, any operation that involves the remote repo implies pkg update (this doesn't always help. One time I forgot I had done pkg upgrade and then came back two days later after a new package set had been deployed), so it always starts from a plausible state. Second, it has a separate fetch and install phase, so will not start installing until it has a consistent snapshot of packages that it needs, so if the package repo changes in the middle, you don't end up having to uninstall packages that were installed but are now the wrong version.