#pulseaudio

[andmaybegayer]

Should I be using pipewire over pulseaudio and if so why?

If you're on Wayland you should definitely at least check out Pipewire, because Pipewire and Wayland both follow a similar philosophy and are kind of being developed side by side. Pipewire with portals is one of the ways to do screen captures and screenshots on Wayland.

If you're running Flatpaks then Pipewire's integration with Flatpak Portals and xdg-desktop-portals more generally will simplify media handling for Flatpaks and generally make running Flatpak media applications more reliable and seamless.

If you use Bluetooth audio, Pipewire has simpler first class support for a wider array of Bluetooth codecs (high bitrate SBC/AptX/LDAC/AAC) and generally simplifies the process of setting up Bluetooth devices exactly the way you want over Pulse.

If you currently fight with running Jack sometimes (or worse, simultaneously running Jack and Pulseaudio) then you should definitely check out Pipewire, because Pipewire implements both Pulse and Jack compatibility layers that are way easier to look after and which can run simultaneously without any fuss.

If you're doing music production with Pulse, Pipewire's pro audio mode might give you some small quality of life improvements by reducing latency and improving inter-program audio links.

If you're doing a lot of live video stuff, especially video involving desktop capture on Wayland, Pipewire can simplify shuttling video around because in addition to handling audio, it handles arbitrary media streams, but you might have this worked out however you're already doing it.

If you are just running a standard X11 desktop and have no problems using Pulseaudio right now, you probably won't notice any change if you switch to Pipewire, especially if you aren't running Flatpaks or Bluetooth. Since Pipewire currently implements a lot of stuff through a Pulseaudio compatible interface, your normal actions with pactl and pavucontrol will continue working transparently or with minimal changes if you do switch.

Installing Pipewire is relatively easy if you don't have any custom pulse configuration. You just have to remove Pulse and install the pipewire, pipewire-alsa and pipewire-pulse packages.

[stevensaus]

Two Ways To Access A Remote PulseAudio Session Over SSH

Despite what you may read elsewhere, there are two ways to control the volume on a remote machine running PulseAudio when you can only access it via SSH. I am not talking about using X forwarding here (which does NOT work), and I do not mean running a whole remote desktop session via VNC or X2Go just to mess with the volume. Nor do you need to run PulseAudio as a system service, or edit PulseAudio's configuration files at all. This is done using command line tools only. Specifically, tmux, pulsemixer, and volumerb. This works on a fresh install of Debian bookworm, where I have not altered any PulseAudio configuration files at all. The system in question has audio two devices - a "regular" audio output and a HDMI output. The Problem Because of some very good complicated security reason , unless you are running a system-wide configuration of PulseAudio (which is not a great idea ), you typically cannot connect to an already running PulseAudio daemon when you SSH into a system. When would this come up? In my particular case so that I can adjust the volume on my media center PC from my laptop with a few simple scripts. It's definitely one of those things not everyone will need, but when you do need it, you need it. The Solution For this, I've done the following (the PulseAudio setup is the default that Debian configured, and works perfectly fine locally): - sudo apt install tmux pulsemixer ruby - wget https://raw.githubusercontent.com/uriel1998/volumerb/master/volume.rb - sudo mv ./volume.rb /usr/local/bin/volume - sudo chmod +x /usr/local/bin/volume - Create a tmux session, have at least one open shell window in that session, and detach from it on the machine whose volume you want to control. Not "connected via SSH", but actually on that machine. The tmux portions are actually extra, but they give you a lot more flexibility. And if you're regularly using SSH and don't know about tmux, you'll be doing yourself a lot of favors by knowing even a little about it. pulsemixer is a fantastic TUI program for handling PulseAudio, and I thoroughly recommend it.

But, as you can see in the first video below, if I just SSH in, pulsemixer returns Failed to connect to pulseaudio. That's also what happens if you try X-forwarding, by the way. However, you can see that - for reasons I do not understand - the volumerb script works out of the box. That script is comparatively klunky compared to pulsemixer, but for some reason it works. The only change from that script's normal behavior is that it only seems to work on whichever sink is set as the "default" sink, not all of the sinks. The interesting part comes next in this second video. You have a tmux session still running on the machine that you just SSH'd into. You can attach to that existing session (tmux attach), and in any windows that were created on the local machine, you can use pulsemixer normally. It attaches to the running PulseAudio session with no difficulty. But again, this workaround only works with windows (and sessions) that were originally created locally. In the video you can see that I create - connected over SSH - a new window. In that new window, I once again cannot connect to the PulseAudio session. But as before, volume.rb still works. So there you go - two ways to control PulseAudio on a remote machine via SSH without running PulseAudio as a system service. I do not understand the specifics of it, nor do I understand it well enough to explain it, but just enough to take seriously the warnings that it is a Very Good Idea to not work around the security here. Featured Image by Gerd Altmann from Pixabay Read the full article

[tom2tec]

CamillaDSP ~ IIR & FIR Engine For Crossovers & Room Correction

A tool to create audio processing pipelines for applications such as active crossovers or room correction. It is written in Rust to benefit from the safety and elegant handling of threading that this language provides. Supported platforms: Linux, macOS, Windows. Audio data is captured from a capture device and sent to a playback device. Alsa, PulseAudio, Jack, Wasapi and CoreAudio are currently…

[tilos-tagebuch]

Für Microsoft wäre Poettering sicherlich ein Gewinn. Das Unternehmen stellt mit WSL erste Unterstützung für Linux-Software bereit. Jedoch ist das Projekt noch nicht sonderlich weit fortgeschritten und hat mit vielen Einschränkungen zu kämpfen. So muss etwa ein Nutzer angemeldet sein, um Software im WSL laufen zu lassen – für Dienste eher suboptimal. Hier könnte Poettering sein Know-how einbringen und die Entwicklung zu einem vollwertigen, eigenständigen Subsystem vorantreiben. Poettering hat mit seinen Projekten wie PulseAudio und systemd erheblich zur Modernisierung von Linux-Systemen beigetragen, ist aber gerade unter Linux-Traditionalisten umstritten.

[nudibutch]

linux devs ily but we REALLY need to switch to wireplumber/pipewire for audio

[movementpolar]

Pulseaudio ubuntu

It's worth noting that Ubuntu 20.04 does not have pulseaudio-dlna in its repositories at all, but it does have pychromecast 4.1.0.Īlso, I didn't add instructions for other Linux distributions because this doesn't seem to work in a virtual machine (even though it's in the same network), so I couldn't test it properly. And since Ubuntu 20.10 has pychromecast 7 in its repositories, that complicates things a bit if you want to use the Python 3 branch.īut I got it to work with the help of a Python 3 branch fork, and this article explains everything to get pulseaudio-dlna to work in Ubuntu 20.10 / Pop!_OS 20.10 with Chromecast devices (it should also work with DLNA devices but like I said, I did not try it). ] This now also works on Ubuntu 20.04, Linux Mint 20.x, Pop!_OS 20.04, etc. Ubuntu 20.10 (and Linux distributions based on it, like Pop!_OS 20.10) does have pulseaudio-dlna in its repositories, but this package does not work properly, at least not with Chromecast devices (I can't test it with DLNA devices - it might work with those). There is, however, a Python 3 branch that you can use, but that too is unmaintained for some time, and it has some issues, for example it's not compatible with the latest pychromecast 7.* (which is what Ubuntu 20.10 has in its repositories). The last pulseaudio-dlna release uses Python 2, which is no longer available in many Linux distributions, which means it can no longer be used on modern Linux distributions. ] The now updated instructions below also work on Ubuntu 20.04, Linux Mint 20.x, Pop!_OS 20.04, etc. This article explains how to install and get pulseaudio-dlna to stream audio from Ubuntu 20.10, Pop_OS! 20.10, and other Linux distributions based on this Ubuntu release, to Chromecast devices. Default values a commented out.Pulseaudio-dlna is a streaming server which allows streaming audio from your Linux computer to a Chromecast or DLNA / UPNP device in the same network, via PulseAudio. # Configuration file for the PulseAudio daemon.

# along with PulseAudio if not, write to the Free Software # You should have received a copy of the GNU Lesser General Public License # General Public License for more details. # MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. # WITHOUT ANY WARRANTY without even the implied warranty of # PulseAudio is distributed in the hope that it will be useful, but # the Free Software Foundation either version 2 of the License, or # it under the terms of the GNU Lesser General Public License as published by # PulseAudio is free software you can redistribute it and/or modify Here's my nf is there something I'm missing or something else to change the sampling rate? I was issuing "sudo service pulseaudio restart" commands between changes, I wasn't sure if that was actually restarting the pulse audio daemon or not so I restarted as well for good meseaure on the last change (to 4000 hz) I think I have it downsampled to 4000 hz with no improvment yet. I guessed that editing /etc/pulse/nf is the way to do this. I was messing with pulseaudio today to get the network features and I've got it working (I have a RTP multicast device, I can see it on the other computer) If I move a stream over to it, I can get it on the the other computer and I can get it on the local machine as well but the sound is VERY choppy.Īccording to the pulse audio the fix is to downsample the audio.

[daemonhxckergrrl]

so i've used linux as a daily driver all my life - kid me was *very* into free stuff i could do with a computer - but I've mostly kept to the 'user-friendly' debian-based distros like ubuntu and mint. I'm probably going to build an arch system next time I need to do a reinstall on my daily driver, but now's not the time. That said, I finally made the switch from pulse to pipewire and I'm losing my mind why did we live like that for so long?

and pulseaudio, despite receiving a ton of backlash for its issues, was already a huge improvement for home users over your options of rawdogging ALSA (exclusive device access so no sharing between applications), or old OSS which lacked Bluetooth support (though did support per-application volume control). New OSS is apparently alright but haven't used it.

ALSA + pulseaudio was my first experience of functioning sound (we tried so hard to get onboard sound working in the mid 00s and it just wasn't).

but pipewire ?? all the power and flexibility of JACK while being simple and integrating into anything supporting pulseaudio. I could never go back !!

glad you made the switch and are enjoying it !! also both arch and void have installers now that still shoe you what they're doing but walk you through it vs doing a manual install. and you can absolutely slap your usual DE and applications on top of it if you want something that Just Works :3

[cant-afford-lobotomy]

my dealer: got some straight gas. this strain is called "UNIX-like operating system” youll be zonked out of your gourd

me: yeah whatever. i don't feel shit

me 5 minutes later: dude i swear the pulseaudio daemon just crashed

my buddy Philip pacing: *no sound*

[pancakeke]

My linux laptop worked fine with bluetooth connecting to my headphones automatically etc up until i walked too far away once in lab and they lost connection, and now, turning off bluetooth system mutes pulseaudio output, but not in any way thats affected by my f1 mute key (which is entirely unrelated to bluetooth audio, since it or the vol up down buttons have no effect on bt volume)

👍

all computers in the world are fucked up because this is how they work ^

[intercal]

yesterday I got in the mail:

arduino nano every

rotary encoders (basically a dial that you can keep turning and doesn't have a start/stop)

plastic knobs that DON'T FIT THOSE ROTARY ENCODERS REEEEEE

some new, different EEPROMs that aren't single-wire and are clock-driven instead

I found an addressable RGB (well, BGR) LED strip that I bought like 6 years ago and hey presto it still works!! back in 2018 I had it hooked up to a Teensy, and the Teensy would take commands over serial to display colors on the LED strip, and I would have a Python program watching pulseaudio on my PC and change the colors based on something that @earthnuker wrote (that I'm trying to decipher again) so that bass would show up as red and high notes would show up as blue. ish. I have a video on tumblr of this from that point but siikr is still down so I can't find it. this is all to say that I bought the Nano for this purpose, or at least prototyping putting this into a compact package so it Just Werks (TM). I am remiss about using an entire breakout board to do this though, I've got a handful of STM32F030 chips sitting around and they may be better suited for this idk. it'll probably be good for a prototype so I can reuse the Arduino at a later date since that sucker was $20 and the STM32 chips are like $1 (but are literally just the microchips that you have to solder on to a PCB yourself)

Rotary encoders are for that dumb Arcblip game that I want to turn into a device. I want to use that as the peripheral to control spinning the little circle around, rather than using buttons to do that. I also discovered that it has a push-to-click functionality too, so I don't even need an external button if I go this route. looks cool, datasheet is horrible. the knobs I bought for the encoder have the D-shaped holes, rather than a knurled circle - they didn't fucking have this on the mouser.com page, you literally had to LOOK AT THE DATASHEET FOR A PIECE OF PLASTIC. I am still salty about this. so they're basically useless, I will take them to canada for @party-slug to use instead.

finally the EEPROMs are 24AA02IDs, they are clock-driven instead of single wire. the AT21CS01 device I was trying to get working previously was really cool but it depended on microsecond precision, some operations literally needed to last 1-2µs maximum which the STM32F030 couldn't quite handle. it felt very loose and I couldn't ever get it to work, so I think I needed a faster processor (also, it would start in "high speed mode" instead of the lower "standard" speed, wtf). so now I'm going to try out using this device where I define how fast the clock goes, it can handle ranges from 100khz-400khz which is a lot easier to manage. it's not quite I2C, but it should be more reliable than the stupid time-and-edge based chip I was using before.

thanks for reading my blog

[strangecloud]

Things I wish I knew before I installed Linux

Flatpak. Flatpak is a software packaging and distribution solution, the main feature of which is containerization. What that means is if you install a Flatpak app, it is isolated from most of your filesystem and processes at runtime. The main advantage is enhanced privacy and security. The main disadvantage is that, if you would like your application to communicate with external apps (like Discord's rich presence) or if you want it to access an external filesystem, like a secondary drive, that's going to be a bit harder and require some workarounds.

Display/Audio servers. Depending on your distro, your server processes and daemons are going to be different. A normal user doesn't have to worry too much about stuff like PulseAudio unless they are already comfortable tinkering anyway, but Display is another matter. If you notice issues with tearing, stutter, or reduced performance, check whether your Desktop Environment uses X11 or Wayland. If one gives you trouble, switch over and try again. There's more advanced tinkering you can do to improve the display experience but I'm not comfortable telling people to root into their xorg.confs or whatever, so just search on the internet if you feel something's not working right.

Ext4. Ext4 is the preferred filesystem format for Linux OSes. There are exceptions but most distros are going to require this format. Importantly, Windows does not read ext4 out of the box. That means if you decide to back up some data onto a default Linux-formatted ext4 drive, you won't be able to access your files from a Windows PC. In this case, you'll need either some data recovery tool or a live Linux boot media to move your files back.

Mesa. Mesa is the open-source graphics library for AMD and Intel graphics adapters. If you use an AMD or Intel GPU (dedicated or not), you'll most likely want to use Mesa and not bother with proprietary drivers unless you are on a workstation use-case. While Linux has a reputation of being friendly to old tech, support for my Legacy GPUs have been mixed. On Debian, my AMD R7 260X does not play nice with RADV and requires a workaround to force AMDVLK. Your mileage may vary.

[tom2tec]

Pipewire ~ Linus Audio Processing

PipeWire is a project that aims to greatly improve handling of audio and video under Linux. It provides a low-latency, graph-based processing engine on top of audio and video devices that can be used to support the use cases currently handled by both PulseAudio and JACK. PipeWire was designed with a powerful security model that makes interacting with audio and video devices from containerized…

[dkplayer]

SDL 3.0 теперь будет предпочитать PipeWire

Для широко используемого слоя программно-аппаратных абстракций SDL, который обычно используется в кроссплатформенных играх, в предстоящем выпуске SDL 3.0 появилась логика, позволяющая предпочесть использование PipeWire напрямую, а не PulseAudio при успешном обнаружении наличия PipeWire.

https://www.gamebuntu.ru/news/sdl-3-0-teper-budet-predpochitat-pipewire/

[linuxlife]

Linux Life Episode 84

Hello everybody and welcome back to Linux Life I admit its been a little while since I made an entry which was informing you of the death of Mangelwurzel and entry of the new Dell Inspiron 3525 I am currently using to type this article on.

Reason its been so long was I actually went back to Windows (stop spitting at back please it’s not pretty) because I wanted to play Blur and a few other games. Well I did and in that time I also tried several ways to have a Linux installation in some form of VM.

So first I used Virtualbox which is a Type 2 Hypervisor basically what that means is everything is emulated in its own little sandbox and all the hardware is virtual. I did manage to get Ubuntu Studio running but to say it was rough would be an understatement.

Virtualbox does not like KDE to say the least and it was painful to run. Also the networking was slow. Sure I could have used normal Ubuntu to see if GNOME was faster but for all I have tried it many times I really don’t like GNOME. Everything needs extensions and the interface is horrible.

Finding apps requires typing in a search because of that stupid Mac Launchpad icon design. I really detest it. I’m sure if you were using a touchscreen then it makes sense but most laptops definitely are not.

Anyway after far too many hours wasted with that I moved on to try Hyper V which is a Type 1 Hypervisor which allows much more connection to base metal components... it lets you use your actual graphics card, the hard drive can be accessed better and you have access to USB and the likes which makes it much better.

Now at the time I was using Windows 11 Home as that’s what came pre-installed on the Dell. So I had to download Hyper V as it wasn’t there by default. So first I try Ubuntu as Canonical have an agreement with Microsoft due to WSL.

Sure enough it installed and was pretty fluid but it had no sound. There was an enhanced mode which refused to work as to do so it needed RDS (Remote Desktop Support) which also prevented sound from working (Huh ?).

To get RDS you need Windows 11 Pro which Microsoft informed me if I wanted a key would be £119. Hmmm not happy obliterated Ubuntu VM. Went web-side got a Windows 11 Pro for £1.67 as it was OEM key. They key would have been £3.35 but I found a 50% off code.

After a bit of fiddling I got Windows 11 Home to go to Pro using a key to switch it then once it did that registered the new OEM key and it worked and guess what I have access to Remote Desktop Services.

So as I felt that Hyper V could cope I now tried installing EndeavourOS which installed without a single hitch. I admit the network was slow but a million times quicker than Virtualbox.

It had no issues with KDE whatsoever but once again no sound. I installed several Hyper V tools and things from the AUR. Even changed out pipewire back to pulseaudio but no matter what I did, sod all worked. RDS couldn’t see the VM even including doing the various Powershell tips.

Bugger all, now sure I could have run Endeavour without sound but I had come this far damn it. OK deleted VM. OK I now have Win 11 Pro so lets try the Ubuntu stuff again.

OK now enhanced mode shows a dialogue and when I log in screen goes black and bugger all happens. Click off enhanced mode and Ubuntu is working. Enhanced mode just black. Setup Ubuntu. reboot now enhanced mode doesn’t come up and even the VM crashes now.

This wasted too many hours and days of messing around. I even got the RDS side running but when I tried to Remotely Access using RDS it crashed. What a complete waste of bloody time.

I even tried running a Ubuntu Studio VM using VMWare Player 17. It didn’t work.

I doing think about setting up a KVM version of QEMU to run a Linux installation but guess what if you want sound on QEMU you need to get a specific version called the Screamer and it won’t let you use KVM setup meaning it wouldn’t allow you to install the Linux.

By this point I just gave up I honestly don’t care what type of Hypervisor it is I am obviously too thick to get things running as they should and I gave up.

I almost tried Proxmox and XCP-NG but upon seeing what they entailed just gave up.

After all that I thought I would leave alone and just use Windows. Then everyone started informing my Facebook had been cloned, well lets just say that didn’t happen when I had Linux.

The internet under Windows just ran slower than when I had Linux. Watching YouTube stalled regularly, downloads failed due to time out and my frustration levels were getting to breaking point.

I tried to dual boot Linux as I found out how to shrink the partition in Windows and use it for Linux installation. So tried EndeavourOS but because of the Secure boot it wouldn’t let the system install.

However for some reason Fedora will install if Secure boot is still on. So I installed Fedora Beta 40 because I wanted to try the new KDE Plasma 6. Now I know it’s beta software so not to have many expectations but most of the software I use was not in the store nor could I find the appropriate dnf installer.

However Fedora now would boot but my Windows 11 wouldn’t the partition was there but it didn’t add it to the menu so now I could load Fedora but Windows wouldn’t load.

Tried OS Prober, Grub Customizer, no matter what I did could not get the Windows 11 partition to boot or get it in the options

Pissed off now I finally turned off Secure boot. Wiped the whole hard drive and installed EndeavourOS Galileo Neo (the latest version) and set up the Dell with KDE.

It now is running on KDE 6 Plasma 6.0.2 running Wayland. It is running seamlessly, the internet is quicker. No issues with YouTube. Updating fine. When I tried KDE 5 Plasma Wayland, Steam and several other programs stopped working.

Well Steam is working fine using XWayland (a Wayland to X11 bridge) and there is a version of Cairo Dock that works with Wayland (it is experimental but so far has not died on me).

So all is right with the world in that respect. Only issue is screen capture isn’t great in Wayland but someone is working on it. OBS and other video capture work fine but screen capture programs such Flameshot, Shutter and other need work.

Well Wayland is slowly being adapted so in time I am sure these things will be fixed over the next few months as more and more main distros seem to be moving to Wayland.

It works well enough and to be honest it's even faster for program launch it seems than the original X11 which is now over twenty years old.

I am yet to try the Steam Proton games such as Path of Exile but who knows maybe next time we speak I may have done so.

So with all that madness abound I bid you farewell for this episode.

Hopefully I will try not to break anything major... until next time take care.

[kennak]

私は今でも個人のデバイスやサーバーで毎日 systemd を使用していますが、私の批判の一部を以下に示します。 * セキュリティが悪い Systemd は、多くのコードが root として実行されるように設計されており、メモリ安全ではない言語でも書かれています。 これは、攻撃対象領域が広いことを意味します (安全性を確保するには、多くのコードにバグがないことを確認する必要があります)。また、コードに本当に深刻なセキュリティ関連のバグがないことを確認するのが困難です ([NSA が推奨]( https ://www.nsa.gov/Press-Room/Press-Releases-Statements/Pre... ) このような重要なことにはメモリセーフ言語を使用します)。 確かに、同様に重要でメモリ安全ではない言語で書かれた他の重要なプロジェクト (この Linux カーネルなど) はありますが、systemd にはいくつかの [かなり重大な脆弱性]( https://cve.mitre.org/cgi-bin/cvekey. cgi?keyword=systemd ) は、この種の危険な言語を安全に使用する能力に自信が持てず、この種の問題を検出して防止するための Linux カーネルとほぼ同じ予算がないことを発見しました。 最新の SELinux を実行している大企業の顧客であれば、これはそれほど問題ではありませんが、現在の機能をすべて保持しながら pid1 の攻撃対象領域を制限する方法で systemd を作成することは可能でした。 * Journalctl はデスクトップ ユーザーや小規模チームにとって面倒です Journalctl は、systemd がログを管理する方法です。 デフォルトでは、追加のメタデータを含むバイナリ形式でログが保存されます。 これにより、大規模なチームがエンタープライズ展開向けに redhat が提供するような集中ログ収集デーモンにログを取り込むことが容易になりますが、古いシステム管理者が使用していた多くのワークフローが中断されます。 大量のログを 1 か所に rsync するだけ、または grep や find などのツールを使用してログを検査するなどです。 もちろん、Systemd はこれらのツールの代替を提供します。grep を使用する代わりに、journalctl を使用してログを検索できますが、grep を使用してテキスト ファイルを検索することもできます。 構成ファイル、ソースコード、またはログ。 ここで、もう 1 つのツールのすべてのフラグを記憶し、ログの収集方法について多くのことを変更する必要があります。 これは、システムから SD カードを取り出すだけではログを確認できないため、組み込み作業を行う人々にとっても課題となります。 ホスト マシン上で動作するjournalctl CLIが必要です。 --directory フラグが追加されましたが、以前はこれがはるかに困難で、ログを読み取るために組み込みシステム (奇妙な方法で壊れる可能性があります) に実際に chroot する必要がありました。 Journalctl には利点もありますが、それらを活用しているのは主に大企業です。 はい、それが実際には systemd-journal または単に「ジャーナル」または彼らがそれを呼んでいるものであることは知っています。 Journalctl は、ほとんどのユーザーに馴染みのあるコマンドです。 * [行動の局所性]( https://htmx.org/essays/locality-of-behaviour/ ) が不十分なため、推論が難しくなります システムがどのように機能するかを理解しようとするとき、すべてを 1 か所で確認できるのは便利です。 systemd ユニット ファイルが存在できる場所は 9 つほどあり、ユニット ファイルにオーバーライドを適用したり、ユニット ファイルはソケット ファイルなどの他のファイルに依存したりできます。 これは、企業内の特定のグループがシステムの一部に対する所有権を主張しやすくなるため、大規模なチームにとっては良いことですが、中小企業で作業するデスクトップ ユーザーやシステム管理者としては、非常に複雑さが増すことになります。 「ls /etc/init.d」と入力するだけで、どのようなサービスが存在するのかの大まかな概要を把握することはできません。さらに systemd 固有のコマンドを覚える必要があります。 サービスを編集したい場合は、単にサービスを編集することはできません。別の systemd 固有のコマンドを使用してオーバーライドを作成し、EDITOR 環境変数が設定されていることを確認してから、別のエディタで元のサービスを開く必要があります。したがって、2つを比較できます。 これにより、作業がさらに複雑になり、systemd 固有のツールを多数使用するようになります (そして、おそらく Red Hat 認定トレーニングを受けることになります)。 * 人々は systemd の準備が整う前にそれを使用します これが systemd や redhat のせいにできるかどうかはわかりませんが、redhat の公式立場では、systemd-resolved は実稼働環境に対応する準備ができていないにもかかわらず、あちこちで使用されているということです。 このため、ネットワーク設定や systemd-resolvd の中断でわずかでも何か違うことを 9 回試みようとすると、systemd-resolvd の多数のセキュリティ問題は言うまでもなく、systemd に対して悪い印象を与える可能性があります。 * Unix の哲学 多くの人は、UNIX の哲学は重要ではない、または systemd は UNIX の哲学を取り入れていると言います。 上で述べたロギングと systemd 固有のツールについてのことですか? これは、UNIX の哲学について語るときに人々が実際に意味していることです。つまり、ログやリモート サーバーへの rsync ログを grep できることを意味しています。 つまり、標準のテキスト ファイルを使用し、ユニット ファイルを「適切に」編集するためにテキスト エディタをラップする特別なコマンドを使用する必要がありません。 * chroot では実行されません 組み込み Linux とサーバー Linux に時間を割いている者として、これは私の個人的な不満にすぎません。 ワークステーションで systemd を使用していない場合、systemd を使用する組み込みシステムをデバッグするのが非常に困難になります。 systemd の使用を強制されているように感じることがあります。私はそれをほとんど克服しましたが、まだ少し苦いです。 また、KoBo 電子リーダー上で完全な Linux ディストリビューションを実行するなど、いくつかの小規模な個人プロジェクトも、必要以上に困難なものになりました。 Docker 内では混乱が生じており、組み込みシステムの大量のテストを行うために、nestybox などの代替 OCI ランタイムを使用する必要があるのはなぜでしょうか。 ありがたいことに、私は Docker の早期導入者ではなかったので、すでに成熟したソリューションが登場するまで、こうした問題に悩まされることはありませんでした。 理由はありません しかし、実際には、pid1 として実行する必要がある 。podman ツールや systemd-nspawn など、systemd と深く統合されている docker の代替手段を使用してほしいということ以外にあります。 これは、独占に近い立場を悪用しようとするあまりにも露骨な試みであったため、余計な不満を抱かざるを得ません。 * OpenRc は systemd が行うすべてのことを実行しますが、それよりも優れています 残念ながら、Gnome のような他の Red Hat の影響を受けたプロジェクトは、デフォルトのサービス ファイルを提供することはおろか、非 systemd init システムをサポートしたりテストしたりしません。つまり、gnome と互換性を持たせたいディストリビューションは、追加の多くの作業を行う必要があります。サービスファイルを作成してテストする作業。 あるプロジェクトの場合はそれが合理的である可能性があり、確かにその追加の作業を行うディストリビューションも存在しますが、可能な限り上流のソースに近づけるという明確な目標を持っている Arch のようなプロジェクトの場合、それは多かれ少なかれ不可能になります。 Systemd が生き残っているのは、それが問題に対する優れた解決策であるからではなく、大企業の支援があり、広く導入されており、コードを作成しても安全であるためです。 IT業界には「IBMを買収したことで解雇された人はいない」という古い格言があります。 安全な業界標準のオプションを選択すれば、たとえそれが技術的に劣ったオプションであっても、失敗しても誰もあなたを責めることはできません。 私は systemd 嫌いではありますが、個人のデバイスやサーバーでは今でも systemd を使用しています。これは、これが最も抵抗の少ない方法だからです。 pulseaudio と Pipewire のような同様の状況が発生することを願っています。pulseaudio の書き換えはオリジナルよりもはるかに優れていました。 ただし、systemd の速度が低下するまではそんなことは起こらないと思いますが、現時点で systemd を再実装しようとすると、規格や動作が変更されて敷物が引き抜かれることになると思います (私は真剣に考えていました)少なくとも比較的単純なユニット ファイルについては、自分で行うことについて)。 動作の局所性の問題に書き換えがどのように対処するかわからないため、私は依然として OpenRc を使用することを好みますが、systemd は時間の経過とともに改善され、信頼性が向上しています。

musl ディストリビューション管理者の目から見た Systemd | ハッカーニュース