Author: Simon Hobson Date: To: dng@lists.dyne.org Subject: Re: [DNG] What does Linus do?
Alessandro Selli <alessandroselli@???> wrote:
>> Do I understand you ccorrectly: that the udev rules are flexible
>> enough to do the right thing, but they are too hard to use?
>
> Yes. On some occations I had to find out where in /sys a device had it's
> control and attribute directory (not easy at all to a newbie) and then run a
> command such as:
>
> # udevadm info --attribute-walk /sys/devices/pci0000\:00/0000\:00\:1a.0/usb1/
>
> to get a long list of attributes to select from when building my own rule
> that isolated a single device.
Unfortunately here we seem to be hitting the complexity vs flexibility tradeoff. AIUI pretty well all attempts to hide the complexity from the user (or admin) end up making things worse - hence the discussion of getting rid of systemd !
> And I remember in a few occasions my old rules
> stopped working and I had to redo them to reflect a change in rules syntax
> that was introduced from a udev's version onwards.
That's not an issue with the rules per se, that's an issue with developers moving the goalposts and not caring what they break. Hmm, and here we come back to the development style of those in charge of systemd again !
> And I could not have udev run a command at device hot-plug. Too cumbersome and little intuitive. And not well documented, at least at the time (a few years ago).
I agree they are not easy (certainly doesn't look it), but then the only involvement I've had is letting "the system" create rules for ethn and changing the name to something I want.