Subject: Re: Bluetooth
To: Matthias Drochner <>
From: Iain Hibbert <>
List: tech-kern
Date: 05/27/2006 22:45:49
On Sat, 27 May 2006, Matthias Drochner wrote:

> -For simple cases of submatch functions we have a
>  "config_stdsubmatch" now. (That's actually my major grief
>  with the USB code - it overloads the "uhub" interface with
>  completely different stuff.)

Ok, though its undocumented (thats my major grief :), will change that..

> -Can you explain why the "bthub" pseudo-device is necessary?

Just because of the abstraction of the bluetooth protocol stack. There are
several layers and although you can always specify a local address, there
is no hard association with any particular controller except at the bottom
level, so it seemed more logical (and very much simpler) to attach a
remote device to a pseudo hub rather than a particular controller..

Although its not quite the same, ppp0 does not attach at com0..

> -Use of "sdp" is namespace pollution IMHO, that word is
>  massively overloaded.

Do you refer to libsdp here? Its true, libsdp could no doubt be included
as part of libbluetooth. Both originated in FreeBSD and while I have
extended libbluetooth (to handle config file parsing*) I didnt do much to
libsdp, was vaguely trying to keep compatibility. I thought about
combining the two at one point, and it should be easy enough.

as to sdpd and sdpquery, not sure how to clear them away - sdp is what
they are, but if you have a better suggestion..? Both again, originated
with the FreeBSD programs - sdpd has only minimal changes (to support the
headset audio gateway) and sdquery (was sdpcontrol) I extended to generate
config entries.

Actually, I'm not so sure that I like the FreeBSD SDP implementation so
much, really it could do with more work I think..


* since the proplib update, it may be better use some kind of xml format
for device configuration, I need to look into that..