[sane-devel] Interactive use of scanner buttons

[Ralph Little]

Hi,

On 2021-12-02 2:35 p.m., Paul Wolneykien wrote:
>    As I can see, XSane doesn't poll the scanner's hardware buttons.
> And if I understand the scanbd manual right, the daemon locks the SANE
> interface to the scanner until a button is pressed, and then gives
> exclusive access to the scanner for the frontend. So, the daemon is
> primarily for launching a frontend, but not for controlling the
> frontend while it is running. Am I right?
>
>    I've grep'ed the mailing list archives a little and it seems that the
> main purpose of various button daemons (scanbuttond, scanbd) is to
> implement such a facility as "push scan": the user goes to a remote
> (possibly, network) scanner, places a standard material into it,
> presses a button on the scanner and the system performes an automatic
> (that's important!) scan.
>    It also seems that in other situations, when a user wants to use an
> interactive frontend (for making previews, selecting the scan
> area, adjusting black and white points and gamma) the scanner is
> expected to be placed on the table near PC. Or, at least, that the
> material for scan can be placed into the scanner without a need for
> further attention (i.e. a sheet material), so the user can set it up
> and then back to his/her keyboard and mouse to make a scan or a number
> of scans with different settings using the GUI frontend.
>
>    These two stories (the "push story" and the "desktop story") are nice
> and simple, but are more or less unapplicable in the world of book
> scanning!
>
>    Firstly, books are all different, so you need to make previews,
> select areas and adjust things for each one. Secondly, books aren't
> flat, so even if you use a special hardware you almost always need
> to handle the book with both your hands!
>
>    So, for scanning books, it would be nice to be able to control such a
> frontend as XSane with the scanner's hardware buttons. Because most of
> the time you can reach the hardware button even if you handle the book
> using both hands (however, a configurable delay after the button is
> pressed may be helpful option).
>
>    (The other possible way to solve the described problem is to use a
> separate triggering device. May be a pedal! But that's another story...)
>
>    I want to know/discuss, what a design of a button monitor for
> use with interactive frontends would fit into SANE better. Any ideas?
>
Button control has always been lacking in SANE and the Linux scanning 
world compared to the Windows counterpart. scanbd is not well documented 
and setting it up is not straightforward.
There are two modes that I can think of, that are actually implemented:

1) Button launch - you need some process to monitor/poll for button 
presses, then launch a pre-programmed application to perform the 
requested function. This would be the scanbd scenario. I have not 
experimented with scanbd myself.

2) Scan on button press - some backends allow you to defer the scan 
function having launched it in some frontend, until a button is pressed. 
The support for this is often provided in the backend and I think this 
is a poor solution. The backend should surely just expose the button 
press functionality to the frontend.

Surely some work could improve this area.

There are also some other considerations that impinge on this aspect. 
Some time ago, I added some functionality to a backend to add support 
for a scanner that has a little LCD window that shows a number which you 
can increment/decrement with some buttons. This might be used to 
indicate a number of print copies perhaps. The backend can now do this, 
but for it to be effective in the frontend, the value has to be 
regularly polled. xsane can show this value in the Advanced window after 
requesting it from the scanner. However, regular polling can be 
detrimental as a general feature. One commentator suggested that over a 
network, this could become an unnecessary bandwidth hog. In SANE we 
would want to add polling features in a generic fashion and we should 
wary of undesirable side-effects. Flooding a network just because you 
are sitting in an idle instance of xsane is certainly such an example. 
It is something that we should discuss carefully before enhancing the 
standard.

I would certainly be happy to see this area improved.

Cheers,
Ralph