[arch-dev-public] Risky business: udev upgrade

[Aaron Griffin]

On Sat, Aug 30, 2008 at 5:40 PM, Aaron Griffin <aaronmgriffin at gmail.com> wrote:
> On Sat, Aug 30, 2008 at 5:23 PM, Tobias Powalowski <t.powa at gmx.de> wrote:
>> Am Sonntag 31 August 2008 schrieb Aaron Griffin:
>>> On Sat, Aug 30, 2008 at 4:28 AM, Xavier <shiningxc at gmail.com> wrote:
>>> > On Sat, Aug 30, 2008 at 1:45 AM, Aaron Griffin <aaronmgriffin at gmail.com>
>> wrote:
>>> >> start_udev is still there because people were jackasses and didn't
>>> >> update initscripts when they updated udev.... or something.. I can't
>>> >> remember the issue, but it was people being foolish and expecting
>>> >> their systems to boot fine.
>>> >>
>>> >> Is everyone ok with removing it?
>>> >
>>> > A suggestion was made 3 times to simply add a conflict line :
>>> > http://bugs.archlinux.org/task/11112#comment31331
>>> >
>>> > This is just a safety against foolish people.
>>>
>>> Added locally.
>>>
>>> Copied from arch-general (whoops, replied to the wrong list) regarding
>>> the readme file we ship with the udev package
>>>
>>> > Actually, I think we should remove this file. Reloading rules and all
>>> > that is covered by the man pages and any arch specific documentation
>>> > should be added to a wiki page so anyone can edit it.
>>> >
>>> > Any issues with removing this? We don't ship custom readme's with any
>>> > other packages that I know of.
>>>
>>> Additionally, do we need the migrate-udev script anymore? I sincerely
>>> hope not. Any qualms with killing THAT off too?
>>
>> migrate udev creates device nodes which might be necessary, i wouldn't kill
>> it.
>
> It creates device nodes that the installer also creates. It only
> appears useful if someone is installing improperly from an existing
> system or from an older ISO. And, hell, if we *really* want to make
> sure these devices exist, we should add checks to rc.sysinit to create
> them before starting udev, rather than relying on a script to do so.

Done http://projects.archlinux.org/?p=initscripts.git;a=commitdiff;h=119b8df1fb1258231750309f01e747e72f382493

Now migrate_udev is safe to remove. This patch makes more sense
anyway, and covers all our bases without forcing the user to run a
script if they screw up.