hal draft spec

[David Zeuthen]

On Wed, 2003-09-10 at 02:19, Havoc Pennington wrote:
> I think your mail looks like it's very much on the right track, fwiw.
> I'd suggest that we create an account for you on freedesktop.org and you
> can start working in CVS and maybe set up a download/information web
> page. A dedicated mailing list might be smart, too, but you're welcome
> to use this one for now. May take a little more time than usual to get
> things set up since we're in a server migration.

Thanks for the feedback. 

So I guess I will create a page with some kind of draft specification of
libhal and hald components and how desktop environments and applications
interact with these - basically based on what was in my initial mail and
any feedback I may receive from this list. 

When that is settled, I need to tweak my prototype implementation to
adhere to that draft spec, put in some (quick and dirty?) device-list
format (a requirement for being useful for anyone but me), and then
release it for public scrutiny :-)

> In any case, if you want send me an ssh v2 public key (ssh-keygen, send
> id_dsa.pub) and a preferred login name and I'll create an account.

Done.

> Once you have a place to work it's really just a matter of making the
> code run, and collecting as much feedback/info as you can from potential
> library users - GNOME/KDE people, distribution people, whoever - I find
> that can be crucial to getting a library accepted, since it's hard to
> anticipate a priori which issues will be important to someone.

I agree - I think the best thing initially is to focus on a) having a
proper spec; b) release an implementation conforming to that spec so
people can try it out; c) getting buy-in to that spec; and finally d)
make timely releases whenever the spec changes.

We might change implementation code later, but that is less important if
desktop application authors / device vendors / os vendors will buy into
the spec. 

I guess the important thing here is to cater for the desktop application
authors initially, since they are capable of writing device-info files
themselves for testing.

> I guess a proof of concept demo showing the desktop nicely handling a
> USB device would do a lot to sell the library as well. ;-) Maybe someone
> will even volunteer to help out with the desktop bits.

That sounds like a plan - I suspect that the aforementioned *_hal_watcher
program is (one of) the desktop pieces missing for such a POC. Any help on 
this is welcome. 

I realize that I need to release some implementation before any real work 
can start on the desktop pieces. I'm hoping to have that ready within the
next few weeks.

  David