stefan11111 wrote:USE="udev" systemd-utils is udev.
Udev is part of systemd that is extracted by the gentoo devs as a standalone package.
Based on the prior quote, I didn't find it worth mentioning. I disagree with your assertion that udev + systemd-utils = udev.stefan11111 wrote:Most people are too shy to comment.
I think the issue has always been no Gentoo developers or users being willing or able to maintain it. Originally the package in it's entirety, but now apparently even maintaining the ebuild. This seems to be a common pattern for various packages, but for semi-obvious reasons, it is mostly seen among those choosing to not use systemd and related software. I include myself among them, but I don't have the skills to do the dev work, so at best I could only do testing. And I'm averse to having my primary systems unavailable, so I'd need to sort out a development environment (which has its own problems). It was roughly around the time Gentoo dropped support for libressl that I admitted to myself that offering "constructive complaints" wasn't useful to anyone, so I try to avoid offering it. It was probably a while before that when I lost most of my interest in, to borrow a phrase, "faffing about" with OSes. I now try to stick with mainstream ::gentoo with few exceptions.stefan11111 wrote:So it is a gentoo issue.

pjp wrote:I disagree with your assertion that udev + systemd-utils = udev.
$ eselect news read 19
2022-04-19-systemd-utils
Title Migration to sys-apps/systemd-utils
Author Mike Gilbert <floppym@gentoo.org>
Posted 2022-04-19
Revision 1
The sys-apps/systemd-utils package was recently added to the gentoo
repository. This replaces sys-apps/systemd-tmpfiles,
sys-boot/systemd-boot, and sys-fs/udev with a single package, and is
for OpenRC users. It does not depend on sys-apps/systemd and contains
the same exact components as the split packages.
USE flags are provided to allow each component to be enabled
or disabled. This change was made to significantly ease maintenance of tools
split out from systemd.
When upgrading to sys-apps/systemd-tmpfiles-250,
sys-apps/systemd-utils[tmpfiles] will be pulled in as a dependency.
When upgrading to sys-boot/systemd-boot-250,
sys-apps/systemd-utils[boot] will be pulled in as a dependency.
When upgrading to sys-fs/udev-250, sys-apps/systemd-utils[udev] will be
pulled in as a dependency.
At a later date, sys-apps/systemd-tmpfiles, sys-boot/systemd-boot, and
sys-fs/udev will be masked for removal once a suitable version of
sys-apps/systemd-utils has been marked stable and sufficient time has
been provided for users to migrate.
Possible problems when upgrading:
1. If sys-fs/eudev is present in the world file (@selected), emerge will
abort the upgrade with a unsolvable blocker error. To resolve this,
either remove sys-fs/eudev from the world file
(emerge --deselect sys-fs/eudev), or disable the 'udev' USE flag for
sys-apps/systemd-utils.
2. The 'boot' USE flag on sys-apps/systemd-utils is disabled by default.
Users migrating from sys-boot/systemd-boot will need to enable the
'boot' USE flag (in package.use) to continue receiving updates.
3. If you have package.use entries for any of sys-apps/systemd-tmpfiles,
sys-boot/systemd-boot, or sys-fs/udev, please update the relevant lines
to refer to sys-apps/systemd-utils instead. This might include
ABI_X86_32 for some users!
It's USE="mdev" sys-apps/busyboxpjp wrote: I don't see an mdev in ::gentoo, so I'm not certain what it is, although it sounds familiar. If it's not in ::gentoo, that's why I'm not using.
As this choice had 0 hits, I made it. That's how I manage /dev on my servers, and they work well.stefan11111 wrote:It's USE="mdev" sys-apps/busybox

In the thread linked by sam_, you say that you need libudev. However, the thread is about a libgudev problem. You can disable udev where everywhere except where you really need it.Syl20 wrote:As this choice had 0 hits, I made it. That's how I manage /dev on my servers, and they work well.stefan11111 wrote:It's USE="mdev" sys-apps/busybox
On my stations... well... However I take the problem, because of my devices (HP printers and scanner, network cards...) and how I use them, I do need a working libudev library.
But, as long as I could resist against systemd stuff, I will do. So, my stations still run eudev, even if it's been (re)masked now. For how much time ?
Sometimes updates become a headache (and sometimes that makes me angry and uselessly aggressive. Again, sorry maintainers)... Today, I got trouble with the eudev package _and_ USE flag masks. This kind of annoyance seems more and more current to me, and that makes me wonder if, in the end, definitely stopping updating my systems would be the best thing to do for my mental health.
Code: Select all
# grep -rE "USE|udev" /etc/portage
/etc/portage/package.mask/udev:sys-fs/udev
/etc/portage/package.mask/udev:=dev-libs/libgudev-238-r1
/etc/portage/profile/use.mask:-eudev
/etc/portage/package.unmask:sys-fs/eudev
/etc/portage/make.conf:USE="X a52 aac acpi alsa ao apng archive audio audiofile audit \
/etc/portage/make.conf: dvdarchive dvdnav dvdr egl elogind encode eudev exif extensions extra \
/etc/portage/make.conf: -thin -tls-heartbeat -udev -webkit -wext -zeroconf"
/etc/portage/package.use/xorg:x11-base/xorg-server elogind udev
/etc/portage/package.use/eudev:sys-fs/eudev static-libs
/etc/portage/package.use/libusb:virtual/libusb udev
/etc/portage/package.use/libusb:dev-libs/libusb udevCode: Select all
# equery d libudev
* These packages depend on libudev:
app-text/calibre-5.44.0-r1 (udisks ? virtual/libudev)
dev-libs/libatasmart-0.19_p5 (virtual/libudev)
dev-libs/libgudev-237-r1 (>=virtual/libudev-199:0/1[abi_x86_64(-)])
dev-libs/libinput-1.23.0 (virtual/libudev)
dev-libs/libusb-1.0.26 (udev ? >=virtual/libudev-208[abi_x86_32(-)?,abi_x86_64(-)?,abi_x86_x32(-)?,abi_mips_n32(-)?,abi_mips_n64(-)?,abi_mips_o32(-)?,abi_s390_32(-)?,abi_s390_64(-)?])
dev-qt/qtgui-5.15.10-r1 (udev ? virtual/libudev)
dev-qt/qtwebengine-5.15.10_p20230623 (virtual/libudev)
media-libs/libcanberra-0.30-r7 (udev ? virtual/libudev[abi_x86_32(-)?,abi_x86_64(-)?,abi_x86_x32(-)?,abi_mips_n32(-)?,abi_mips_n64(-)?,abi_mips_o32(-)?,abi_s390_32(-)?,abi_s390_64(-)?])
media-libs/libsdl2-2.28.1 (udev ? >=virtual/libudev-208[abi_x86_32(-)?,abi_x86_64(-)?,abi_x86_x32(-)?,abi_mips_n32(-)?,abi_mips_n64(-)?,abi_mips_o32(-)?,abi_s390_32(-)?,abi_s390_64(-)?])
media-libs/libv4l-1.22.1 (dvb ? virtual/libudev[abi_x86_32(-)?,abi_x86_64(-)?,abi_x86_x32(-)?,abi_mips_n32(-)?,abi_mips_n64(-)?,abi_mips_o32(-)?,abi_s390_32(-)?,abi_s390_64(-)?])
media-plugins/gst-plugins-vaapi-1.20.6 (drm ? >=virtual/libudev-208[abi_x86_32(-)?,abi_x86_64(-)?,abi_x86_x32(-)?,abi_mips_n32(-)?,abi_mips_n64(-)?,abi_mips_o32(-)?,abi_s390_32(-)?,abi_s390_64(-)?])
media-video/pipewire-0.3.77-r2 (virtual/libudev[abi_x86_32(-)?,abi_x86_64(-)?,abi_x86_x32(-)?,abi_mips_n32(-)?,abi_mips_n64(-)?,abi_mips_o32(-)?,abi_s390_32(-)?,abi_s390_64(-)?])
net-misc/freerdp-2.10.0-r3 (usb ? virtual/libudev:0)
net-misc/networkmanager-1.42.6-r1 (>=virtual/libudev-175[abi_x86_32(-)?,abi_x86_64(-)?,abi_x86_x32(-)?,abi_mips_n32(-)?,abi_mips_n64(-)?,abi_mips_o32(-)?,abi_s390_32(-)?,abi_s390_64(-)?])
sys-apps/pciutils-3.10.0 (udev ? >=virtual/libudev-208[abi_x86_32(-)?,abi_x86_64(-)?,abi_x86_x32(-)?,abi_mips_n32(-)?,abi_mips_n64(-)?,abi_mips_o32(-)?,abi_s390_32(-)?,abi_s390_64(-)?])
sys-apps/usbutils-015 (virtual/libudev)
sys-apps/util-linux-2.38.1-r2 (udev ? virtual/libudev)
sys-auth/elogind-246.10-r3 (virtual/libudev)
sys-fs/cryptsetup-2.6.1 (udev ? virtual/libudev)
sys-fs/lvm2-2.03.21-r1 (udev ? virtual/libudev)
x11-base/xorg-server-21.1.8-r2 (udev ? virtual/libudev)
x11-drivers/xf86-video-intel-2.99.917_p20230201 (udev ? virtual/libudev)
Code: Select all
# equery d libgudev
* These packages depend on libgudev:
gnome-base/gvfs-1.50.6 (udev ? >=dev-libs/libgudev-147)
gnome-extra/nm-applet-1.32.0 (>=dev-libs/libgudev-147)
media-gfx/gimp-2.10.34-r2 (udev ? dev-libs/libgudev)
media-libs/clutter-1.26.4-r1 (egl ? >=dev-libs/libgudev-136)
media-libs/clutter-gst-3.0.27-r2 (udev ? dev-libs/libgudev)
media-libs/gst-plugins-base-1.20.6 (gbm ? >=dev-libs/libgudev-147[abi_x86_32(-)?,abi_x86_64(-)?,abi_x86_x32(-)?,abi_mips_n32(-)?,abi_mips_n64(-)?,abi_mips_o32(-)?,abi_s390_32(-)?,abi_s390_64(-)?])
media-plugins/gst-plugins-v4l2-1.20.6 (udev ? >=dev-libs/libgudev-208[abi_x86_32(-)?,abi_x86_64(-)?,abi_x86_x32(-)?,abi_mips_n32(-)?,abi_mips_n64(-)?,abi_mips_o32(-)?,abi_s390_32(-)?,abi_s390_64(-)?])
net-libs/libqmi-1.30.8 (>=dev-libs/libgudev-232)
net-misc/modemmanager-1.18.12 (udev ? >=dev-libs/libgudev-232)
sys-fs/udisks-2.9.4-r3 (daemon ? >=dev-libs/libgudev-165)
sys-power/upower-0.99.20 (kernel_linux ? >=dev-libs/libgudev-236)
xfce-base/libxfce4ui-4.18.4 (system-info ? dev-libs/libgudev)
xfce-base/thunar-4.18.6 (udisks ? dev-libs/libgudev)
xfce-base/thunar-volman-4.18.0 (dev-libs/libgudev)
Code: Select all
gnome-extra/nm-applet-1.32.0 (>=dev-libs/libgudev-147)
net-libs/libqmi-1.30.8 (>=dev-libs/libgudev-232)
xfce-base/thunar-volman-4.18.0 (dev-libs/libgudev)stefan11111 wrote:Do you need them?
Code: Select all
gnome-extra/nm-applet-1.32.0 (>=dev-libs/libgudev-147)Code: Select all
net-libs/libqmi-1.30.8 (>=dev-libs/libgudev-232)Code: Select all
# equery d libqmi
* These packages depend on libqmi:
net-misc/modemmanager-1.18.12 (qmi ? >=net-libs/libqmi-1.30.8[qrtr?])
# equery -N u modemmanager|grep qmi
+ + qmi : Enable support for the QMI modem protocol used by devices with Qualcomm chipsetsCode: Select all
xfce-base/thunar-volman-4.18.0 (dev-libs/libgudev)Code: Select all
# equery d thunar-volman
* These packages depend on thunar-volman:
xfce-base/xfce4-meta-4.18 (>=xfce-base/thunar-volman-4.18.0)
Do you need network-manager, or that applet?Syl20 wrote:stefan11111 wrote:Do you need them?Yes. I often need to switch quickly from a network to another (this computer is a laptop).Code: Select all
gnome-extra/nm-applet-1.32.0 (>=dev-libs/libgudev-147)
You mean usb tethering? no need for that package. All usb tethering needs is kernel support.Syl20 wrote:Well... good question.Code: Select all
net-libs/libqmi-1.30.8 (>=dev-libs/libgudev-232)So, yes. I sometimes need an internet connection via my phone.Code: Select all
# equery d libqmi * These packages depend on libqmi: net-misc/modemmanager-1.18.12 (qmi ? >=net-libs/libqmi-1.30.8[qrtr?]) # equery -N u modemmanager|grep qmi + + qmi : Enable support for the QMI modem protocol used by devices with Qualcomm chipsets
You can just use package.provided if you prefer.Syl20 wrote:I think no. But as there are two really needed packages above, I think I can wait before replacing this meta-package by its children one by one :Code: Select all
xfce-base/thunar-volman-4.18.0 (dev-libs/libgudev)Code: Select all
# equery d thunar-volman * These packages depend on thunar-volman: xfce-base/xfce4-meta-4.18 (>=xfce-base/thunar-volman-4.18.0)
I need both. When I say I often need to switch _quickly_ from a network to another, that means "in two clicks".stefan11111 wrote:Do you need network-manager, or that applet?
OK. Interresting. I'll do a try.stefan11111 wrote:You mean usb tethering? no need for that package. All usb tethering needs is kernel support.]
You're right ! Thank you.stefan11111 wrote:You can just use package.provided if you prefer.


Syl20 wrote:I need both. When I say I often need to switch _quickly_ from a network to another, that means "in two clicks".stefan11111 wrote:Do you need network-manager, or that applet?
Code: Select all
gnome-extra/nm-applet-1.32.0 (>=dev-libs/libgudev-147)
AFAIK, the (e)udev init scripts stop the boot process if DEVTMPFS isn't in use.NeddySeagoon wrote:stefan11111,
(e)udev and friends manage symlinks and permissions in /dev.
I don't think that requires /dev in the kernels DEVTMPFS.
Dynamic adding and removal of /dev entries is performed by the kernel. (e)udev does not add or delete /dev nodes.
Even with a static /dev, you can look in /sys/dev/* to see what dev nodes the kernel would have created and discover the major:minor device numbers.

In that case, your problem is solved, isn't it?Syl20 wrote:Of course not.stefan11111 wrote:That really doesn't need the new libgudev.Already done.Add an older ebuild to your overlay and mask everything newer.

It that case, let's take a look at libudev too.Syl20 wrote:I don't think so. At the beginning, the problem is not libgudev (with the "g"). The problem is that more and more basic components depend on a working libudev (without the "g"), and the lonely gentoo package that provides it today is the systemd-ized version.
That said, thank you to help me to sanitize my system.
Code: Select all
$ eix -I udev
No matches foundCode: Select all
dev-libs/libatasmart-0.19_p5 (virtual/libudev)
dev-libs/libgudev-237-r1 (>=virtual/libudev-199:0/1[abi_x86_64(-)])
dev-libs/libinput-1.23.0 (virtual/libudev)
dev-qt/qtwebengine-5.15.10_p20230623 (virtual/libudev)
media-video/pipewire-0.3.77-r2 (virtual/libudev[abi_x86_32(-)?,abi_x86_64(-)?,abi_x86_x32(-)?,abi_mips_n32(-)?,abi_mips_n64(-)?,abi_mips_o32(-)?,abi_s390_32(-)?,abi_s390_64(-)?])
net-misc/networkmanager-1.42.6-r1 (>=virtual/libudev-175[abi_x86_32(-)?,abi_x86_64(-)?,abi_x86_x32(-)?,abi_mips_n32(-)?,abi_mips_n64(-)?,abi_mips_o32(-)?,abi_s390_32(-)?,abi_s390_64(-)?])
sys-apps/usbutils-015 (virtual/libudev)
sys-auth/elogind-246.10-r3 (virtual/libudev)
Did anything change since 2022 in regards to this?pjp wrote:One thing we can agree on, that is a news item from 2022. I expected a response of that kind or similar, which is why I didn't originally respond after selecting "Something else."stefan11111 wrote:$ eselect news read 19
2022-04-19-systemd-utils
Seriously ? Do you really want me to show that each of these components is really mandatory ?stefan11111 wrote:I understand that you must use networkmanager, which needs libudev, and it's gui needs libgudev.
What do you need pipewire for?
What do you need elogind for?
What do you need libinput for?
What do you need qtwebengine for?
What do you need usbutils for?
What do you need libatasmart for?

I asked in the hopes to maybe trim down that list.Syl20 wrote:Seriously ? Do you really want me to show that each of these components is really mandatory ?stefan11111 wrote:I understand that you must use networkmanager, which needs libudev, and it's gui needs libgudev.
What do you need pipewire for?
What do you need elogind for?
What do you need libinput for?
What do you need qtwebengine for?
What do you need usbutils for?
What do you need libatasmart for?
...
Well, okay.
Your call. I only use alsaSyl20 wrote: 1/ Today, using daily a workstation without pulse/pipewire is a PITA.
I thought you use it for xorg only. If you need it, keep it.Syl20 wrote: 2/ Today, having a little control on what several users can simultaneously do or not on a workstation without a login service is a PITA.
What do you mean "manage input devices"?Syl20 wrote: 3/ What else do you suggest to manage my input devices ?
4/ Considering the loooong time this ugly package takes to compile, do you really think I never tried to avoid it ?
5/ see 3/
6/ see 3/, 5/, and Neddy"s signature.
After several years of hellish (see my previous posts if you have such time to loose) and hard resistance, I finally decided to install pulseaudio to solve my audio problems. I wish you never have to deal with a dmix bug, or sort of.stefan11111 wrote:Your call. I only use alsa
I'm surprised to learn that xf86-input-mouse and xf86-input-keyboard drivers are still maintained upstream. That said, even their maintainers recommend using something else for linux users. Like (now old) xf86-input-evdev, or... xf86-input-libinput.stefan11111 wrote:What do you mean "manage input devices"?
If you mean xorg drivers, use the drivers from neddy's overlay.
https://www.slicer.orgstefan11111 wrote:What do you need qtwebengine for? You didn't answer that. If you don't want to, that's fine.
I hesitate to reply in a heavily derailed thread, but I haven't found a better way to share my thoughts, and this is the only sane message in a topic that doesn't do anything but derail everywhere I read about it, so I wanted to honor it.Hu wrote:For those opposed to running udev, what is the motivation? Are there specific technical problems with the released udev we use today? Is this a philosophical objection since the systemd project hosts the udev code, and systemd's most active contributors have a history of questionable decisions, thereby tainting all projects in that repository? Is there a concern that an announced upcoming systemd-udevd change will introduce technical regressions relative to the currently released udev?