[sane-devel] Epson Perfection1640: Testing focus (TPU)

[ael]

On Sun, Nov 07, 2021 at 08:16:42PM +0100, Wolfram Sang wrote:
> 
> It is true that the option is not a boolean anymore but an integer. But
> still, if you selet the TPU the old default value for 2.5mm above glass
> (89) should be automatically selected. Without TPU, it should use the
> default of on glass (64).
 .. [snip]..
> Please check if that works on your system as well.

I have now tested with dev->focusSupport = SANE_TRUE and the TPU
installed. Again I set the focus to either 64 (on glass) or 89
(above glass).

As with the flatbed case, I could detect no visual difference in the scans.
They were not identical, but then I expect some noise in the system,
so that is not surprising.

My conclusion is that while the scanner accepts (and reports) the focus
as 64 or 89, it appears to make no difference. As before, I strongly
suspect that the scanner itself is doing some sort of internal
autofocus.

So after all that, I think the existing 
dev->focusSupport = SANE_FALSE

is correct so the epson2 backend doesn't need to offer focus control.

It was confusing because that is what it used to do.

In this testing, I hit another bug, not for the first time.

When using xsane to acquire a large scan at bit depth 14,
only a small part of the image was written, and the xsane locked up.
The scanner ready light stayed off. Obviously my first thought was
a buffer overflow somewhere.

I will try a similar scan with scanimage: if that also fails,
I guess the epson2 backend is likely to be the culprit. Time
permitting, I will try and capture debug logs in that case.

ael