Re: gnome 2.19 schedule.

[Richard Hughes <hughsient gmail com>]

On Sun, 2007-03-18 at 13:41 +0100, �ienne Bersac wrote:
> > how "important" is imcapd and at which state is it currently (would
> this
> > introduce an external dependency to imcapd if gnomescan would be
> > included into GNOME 2.20? if imcapd isn't good enough yet, could
> button
> > support just become "experimental", or postponed to 2.22? the same
> > questions for Gnome OCR.)
> 
> Luckily, i can get around imcapd because of the new gnome-scan backend
> handling (very similar to GtkPrint one).

Why do we need imcapd? The thought of yet another service just to
monitor scanner buttons scares me a little. Why not just write a HAL
addon (100 lines of C) to query the scanner? Each type of scanner (hp,
epson, mustek) can have it's own addon, and just have a single abstract
event: ButtonPressed 'scanner' which can be easily handled by the
session.

Addons are launched automatically only if the hardware is connected, all
communication is via audited secure code, and over DBUS. Win win win.

This is how the power buttons are handled in g-p-m, where we have to
deal with APM, ACPI and PMU as very differently structured backends.

This would also mean a small patch to gnome-volume-manager to have an
option "Run this program when the scanner button is pressed", which
again, is only a 100 or so lines of code.

People shouldn't be expected to start a service when they insert a
scanner - this stuff should just work from a users perspective.

Richard.