[sane-devel] hp5590: Scan aborts when using default page size

[Olaf Meeuwissen]

Rainer Keller writes:

> Hi,
>
>> The SANE API only provides for SANE_UNIT_MM for lengths.  Just looking
>> at your patch, the *original* code seems to be correct.  The real bug is
>> somewhere else in the hp5590 backend.
>
> Ok, thanks for the hint.
> Given that all lengths are in millimeters, isn't multiplying mm with dpi wrong 
> then? In this backend there are some places where this happens:
>
> e.g.  backend/hp5590.c:1053
>   scanner->tl_x * scanner->dpi,
>   scanner->tl_y * scanner->dpi,
>   (scanner->br_x - scanner->tl_x) * scanner->dpi,
>   (scanner->br_y - scanner->tl_y) * scanner->dpi,

OK, so I had a look at hp5590.c rather than just your patch.  Looks like
your patch was good after all.

I'd rashly assumed that the backend kept the values of scanner->tl_x and
friends in the same units as the SANE API spec provides.  Turns out that
that is *not* the case.  It keeps these in inches throughout the code so
the calculation you quote above are fine.  The backend's SANE options
convert from and to SANE_UNIT_MM (in sane_control_option) when accessing
these values so the backend user only sees SANE_UNIT_MM values.

The range_x and range_y values are part of the SANE option and have to
be defined in SANE_UNIT_MM.  This discrepancy threw me off, and did the
same with however implemented the ADF support apparently ;-|

Hope this helps (and sorry for the confusion),
-- 
Olaf Meeuwissen, LPIC-2           FLOSS Engineer -- AVASYS CORPORATION
FSF Associate Member #1962               Help support software freedom
                 http://www.fsf.org/jf?referrer=1962