[Hwdb-devel] debian hwdb

Roland Stigge stigge@debian.org
Sun, 13 Jun 2004 15:48:34 +0200


Hi,

On Thu, 2004-06-10 at 09:26, Zenaan Harkness wrote:
> OK, I just checked out the tar ball.

Did you actually look inside it? I'm asking because ...

> There are three classes of users:
> 
> 1) end users
> 2) manufacturers
> 3) driver developers (at least those independant from class 2)
> 
> A debian specific hwdb solves the problem for Debian users in category
> 1) only.
> 
> How limited is that (rhetorical!)!?!

My first draft includes the following user DB user classes: Community
(end users), Manufacturer (e.g. certifying entities), Developers (DDs).
And this isn't limited, IMHO. And no problem to extend it further.

A problem I recognize points in another direction. Since the DB schema I
proposed already looks more complicated than the HWDBs currently in use
by "the competition", I wonder if it is possibly already
over-engineered. Unfortunately, I don't have much feedback at all, yet.
And also, don't have that much spare time left.

It's even no problem to support other distribution releases besides
Debian main (look at the schema that associates supported drivers with
an actual release/distribution). But the question is: Shouldn't we focus
on Debian? Other distributors may have other conceptions of how a HWDB
should look like...

Hint: A UML representation of the current schema is in hwdb.dia. The
SQL-implementation is in hwdb_create. The Python scripts HWDB.py and
db.py represent the framework and the (yet unfinished) web frontend,
respectively.

> The problem for end users is knowing what's supported. They might be
> using GNU/Linux, want to try Debian GNU/HURD, or be on *BSD.

What should be our goal? To advertise the >60 supported
pseudo-architectures the NetBSD folks are so proud of? Furthermore, the
logical next step would be to include information about all kinds of
operating systems. (When Debian currently considers some parts of "free"
OSs/kernels non-free, why not also support Windows, SCO UnixWare etc.)

But feel free to contribute development resources or feedback. Maybe we
come up with a solution everyone's comfortable with.

Thanks in advance.

bye,
  Roland