FDI merging framework

[Richard Hughes]

On Fri, 2008-08-08 at 10:02 +0200, Kay Sievers wrote:
> I expect the subsystem daemons will implement their own
> matching/config/setup logic. A lot of the current fdi functionality,
> like device-classification and global device settings, can just move
> down to udev rules, where they fit better into todays model, and the
> subsystem specific data will just live in the subsystem daemons, to
> fit their specific need. I do not think, that besides udev rules,
> which will provide simple KEY=value environment keys along with the
> event, we will be able to share too many things between the different
> subsystems at this level.

Sure, makes sense. Most of the trivial rules can be shipped as udev
rules -- but I wanted to know if it makes sense to have some sort of XML
high level rule to low level udev rule converter.

> For the -power case, you need to store your history of events and
> measured values somewhere, much like dk-disks stores SMART data and
> such, in its own database. It does not really make sense to try to put
> that into some globaI unified thing. So I think, we should not try
> again to come up again with any abstract model, which could possibly
> be shared between subsystems - in reality it likely causes more
> problems than is solves.

Sure, I'm just dumping the stats values to a flat file right now, I
might switch to sqlite if the amount of data gets huge.

Richard.