David Brownell wrote:
> > > Eventually, it'd make
> > > more sense not to compile specific device IDs into drivers.
> >
> > Why?
>
> Because the original driver has no strong reason to know, or care,
> about every device that's going to be compatible enough to use it.
It is at least usefull to be able to tell the user of a driver (a human or a
daemon) which devices it is *known* to work with.
> > Adding support for a device in a driver will at least force you to recompile
> > the driver, so you do not create extra work. It would also automatically
> > update your bindings when you upgrade the driver.
>
> That's fine for the bindings that are compiled-in. What about for
> cases where a new device is compatible with another, and someone
> just wants to spend thirty seconds getting it up (hack the admin
> database) rather than tracking down the latest/buggiest driver or
> figuring out how to modify that driver herself (or himself)?
This can be easily provided with a little utility that can add entries to the
daemon's database. This is not something that is needed for general use, so
this can be a 'debug' feature.
We could also have a file containing extra configuration data, but I don't
consider this a good idea. It opens the door too much for possible
misconfigurations and inconsistencies. It might be nescessary though to make
evolution easier.
> > true, but what I meant to suggest was that you don't have to adjust config
> > files of the deamon to let it know which driver to load. It can figure that
> > out itself.
>
> True, that can be handy. But in engineering terms, we've got shell
> scripts working already, and something based on specialized symbol
> table manipulation (including getting drivers to adopt that convention,
> which sounds 2.5-ish) isn't yet written. (And if it gets written, it
> could quite easily generate more up-to-date shell scripts ...)
Yeah, I realize what I suggest would impact about every subsystem and every
module in the kernel. It is not a good idea to do this for 2.4 :-)
I was planning a bit more research in what is available/feasable/nescessary
and then write up a serious proposal. Consider this a brainstorm session :-)
> At some level, it doesn't matter how the meta-data (what drivers
> handle what devices) is managed ... except that it be easily changed
> by administrators, and work well with the driver loading framework.
... and not to mention it should be as foolproof as possible to uneducated
users without sacrificing real functionality. And I think this is a point
where my proposal scores high :-)
J.
--
________
/---------------------------------\ /--------------------\
| Johan Verrept \ / Alcatel |
| Software Development Engineer \ / De Villermontstr 38|
| xDSL Customer Premises Equipment \/ B-2550 Kontich |
| http://www.alcateldsl.com/ ALC/\TEL Belgium |
| mailto:[EMAIL PROTECTED] Phone: +32 3 450 3506 |
\-------------------------------------------------------------/
---------------------------------------------------------------------
To unsubscribe, e-mail: [EMAIL PROTECTED]
For additional commands, e-mail: [EMAIL PROTECTED]
