Re: gnome 2.19 schedule.

[David Zeuthen <david fubar dk>]

On Sun, 2007-03-18 at 20:16 +0100, �ienne Bersac wrote:
> Hi,
> 
> This is a true problem i don't really know how to solve. Luckily, SANE
> people have just commited fdi generator to SANE CVS. So we should have
> proper device detection in the next release. 

Right, it was actually committed today

 http://lists.alioth.debian.org/pipermail/sane-devel/2007-March/018764.html
 http://lists.alioth.debian.org/pipermail/sane-devel/2007-March/018763.html
 http://lists.alioth.debian.org/pipermail/sane-devel/2007-March/018851.html

I'll add the docs + other things to HAL so the 'scanner' capability is
defined. Note that HPLip etc. needs some work too since they are their
own little HAL-ish implementation... (I'll spare you for my thoughts on
HPLip - they aren't nice)

> However, this does not
> solve sharing and button event handling.
> 
> The problem with addon is : to bind to SANE or not to bind ? SANE has
> already several backends and tons of device supported. Including buttons
> handling. Imho, writing several backends just for button handling is
> waste.
> 
> However, the problem with linking to SANE is the locking of device. If
> HAL own the device, xsane, kooka, gnome-scan won't be able to use it.
> 
> scannerbuttonsd from SANE CVS use libusb for buttons handling, would be
> nice to have it as an hal add-on.

Yeah. Suppose we have a halified scannerbuttonsd (e.g. running as an
addon, emitting D-Bus signals etc.), would it know to give up access to
the device (e.g. close the file descriptor) when libsane wants to use
it?

     David