udev problems

DJ Lucas dj at lucasit.com
Sat May 22 10:50:40 PDT 2004


Kevin P. Fleming wrote:

> DJ Lucas wrote:
> 
>> Optimized = No.  Another proggy running in the backround when a static 
>> device would suffice.
> 
> 
> FUD. Please research what you are talking about before making statements 
> like that. udev does not run in the background, it runs at startup to 
> initially populate the /dev directory. From that point forward, if you 
> do not cause any hotplug events to occur, there is no difference between 
> it and a static /dev system. If you do cause hotplug events to occur, 
> udev is invoked to handle those events.

Correct...  for udev, but this wasn't about udev specifically.  I still 
should have worded that better.  I know not running in the backround all 
the time (udev), but when a device is plugged in, a script is run to 
create the device node.  devfsd OTOH don't know don't care...  Anyways, 
I *want* udev in the book from both a personal POV and a technical POV. 
  The problem is that I can't justify the change given the books stated 
goals.

I need to look at the wiki's goals too....

Control = Depends....it's optional, that is for certain so it is a 
choice and therefore the user still has controll.

Compact = No.  Same reason as before.

Secure = No.  Yet another process to look after.

Optimized = Again, this could be looked at two ways.  Is optimized 
having the system do things for you, or having a staitic device already 
created and ready to go for you as soon as you need it (coincidently, 
this is the way kernel handles modules IIR/UC).  It's much easier.

Unfortunately here is the big kicker:

> The primary focus in LFS is on learning about the system.

Education = No.  You simply learn more by manually creating the devices 
needed.  But using that argument alone, we could put the mknod commands 
directly into the book and make it *really* educational. ;-)

-- DJ Lucas




More information about the lfs-dev mailing list