[RFC] Proof-of-concept for netlink listener for mdev -i

[Timo Teras]

Hi,

On Sun, 15 Mar 2015 21:46:42 +0100
Harald Becker <ralda at gmx.de> wrote:

> > In theory, the netlink socket (or named pipe) could be set up of a
> > separate process. That way we avoid the init code in memory of
> > longtime running process. Not sure its worth it thoug.  
> 
> You mean splitting the netlink creator, from the netlink waiter, from 
> the netlink reader ... even if I see, it could be possible, for which 
> result? Complexity! ... I'm trying to go the opposite way, clarity
> and modularity. Not only a solution for one group of people.

My take on this is that you are designing an abstract server to be
usable by several things. However, in our case it would be used only by
one thing. Thus adding complexity by introducing one more component
than needed. Not everyone wants to use fifo supervisor.

Yes, it might make sense if everyone used it. But who's now trying to
make people do things his way?

I would happier with solution to a specific problem that is simple to
install, than solution to all that requires additional changes
everywhere.

Granted the idea is nice. But what happened with inetd? No one uses it
real servers anymore. It is usable only in special embedded solutions.
I see fifosvc the same. It'd be usable in special embedded solutions,
but not in a real distribution. The idea of it is intriguing, though.

I have not followed all the discussion. But does it handle also all the
named pipe details correctly? E.g. when named pipe has been connected,
and then disconnected by reader. All writers will get errors like EPIPE
until you close and recreate it.

Just my 5 cents.

/Timo