Hi,
On Thursday 29 September 2005 20:34, Henning Meier-Geinitz wrote:
> Hi,
>
> On Thu, Sep 29, 2005 at 05:05:18PM +0200, Gerhard Jaeger wrote:
> > please find attached a patch to make the OpticSlim M12 scanner
> > work. Work here means, scan a sheet.
>
> I have included the patch into CVS with some modifications. Please
> check if it still works. See below for details.
>
> > Calibration is not available,
>
> How does that work with sheet-fed scanners at all? Is there also some
> kind of calibration area?
as Gerard already stated. there's an extra calibration sheed, which
needs to be scanned at the first time, then the settings are stored
somewhere.
>
> > area selection is limited - positioning does currently not work.
>
> That means that length and width can be changed but not tl-x and tl-y?
Yes and No. Need to do some more investigations on how to set tl-y, tl-x
seems to be fine.
> > The patch is meant as a starting point to make gt68xx based sheet-fed
> > scanner
> > work. It should apply to the current cvs.
> >
> > @Henning: It's up to you to include it. Maybe some changes are needed - I'll
> > test them.
> >
> > Ciao,
> > Gerhard
>
> > --- sane-backends/doc/descriptions/gt68xx.desc.orig 2005-09-27
> > 11:19:02.000000000 +0200
> > +++ sane-backends/doc/descriptions/gt68xx.desc 2005-09-29
> > 16:57:39.000000000 +0200
>
> [...]
>
> > + if (s->dev->model->flags & GT68XX_FLAG_SHEET_FED)
> > + {
> > + s->val[OPT_BACKTRACK_LINES].w = 0x3f;
> > + DISABLE (OPT_BACKTRACK_LINES);
> > + }
> > +
>
> Is it really necessary to disable that option unconditionally?
> Otherwise I would move the check for GT68XX_FLAG_SHEET_FED up to the
> normal setting of that option (normally 0x3f is used for CCD).
Hmmm, probably not.
> > - RIE (gt68xx_scanner_calibrate (s, &scan_request));
> > + if ( !(s->dev->model->flags & GT68XX_FLAG_SHEET_FED)) {
> > + RIE (gt68xx_scanner_calibrate (s, &scan_request));
> > + }
>
> I haven't changed that because coarse and fine calibration is turned
> off anyway for that scanner.
Okay!
> > --- sane-backends/backend/gt68xx_devices.c.orig 2005-09-27
> > 11:18:57.000000000 +0200
> > +++ sane-backends/backend/gt68xx_devices.c 2005-09-29 11:29:05.000000000
> > +0200
>
> I used a new command_set for sheet-fed types. this way we have more
> flexibility.
Tried that in my first approach, but failed - anyway I'll test the changes.
> Also there is one more level of indirection (always call
> gt68xx_device_*, not gt68xx_generic_* or gt68xx_gt6816_* directly).
Okay, I see.
>
> > --- sane-backends/backend/gt68xx_generic.c.orig 2005-09-27
> > 11:18:57.000000000 +0200
> > +++ sane-backends/backend/gt68xx_generic.c 2005-09-29 11:29:49.000000000
> > +0200
> > @@ -96,10 +96,17 @@ gt68xx_generic_read_scanned_data (GT68xx
> >
> > RIE (gt68xx_device_req (dev, req, req));
> >
> > - if (req[0] == 0)
> > - *ready = SANE_TRUE;
> > + *ready = SANE_FALSE;
> > + if (dev->model->flags & GT68XX_FLAG_SHEET_FED)
> > + {
> > + if (req[0] == 0 && req[1] == 0x35 )
> > + *ready = SANE_TRUE;
> > + }
>
> Is this really necessary? I know that 95% of the gt68xx scanners
> return 00 35 but I only test for 00 beacuse of the 5% which don't
> return the 35. If it also works with the test only for the 00, I would
> prefer to keep the old version. At the moment, i have apllied your
> patch, however.
Maybe it is not necessary - I'll test that too.
>
> in gt68xx_geneirc_set_exposure-time:
>
> > + if (dev->model->flags & GT68XX_FLAG_SHEET_FED)
> > + return SANE_STATUS_GOOD;
> > +
>
> Not applied. Added a check in gt68xx_high.c instead.
Okay.
> > --- sane-backends/backend/gt68xx_gt6816.c.orig 2005-06-01
> > 13:28:18.000000000 +0200
> > +++ sane-backends/backend/gt68xx_gt6816.c 2005-09-29 11:23:28.000000000
> > +0200
> > @@ -217,6 +217,9 @@ gt6816_carriage_home (GT68xx_Device * de
> > {
> > GT68xx_Packet req;
> >
> > + if (dev->model->flags & GT68XX_FLAG_SHEET_FED)
> > + return SANE_STATUS_GOOD;
> > +
>
> Not applied. Added checks in gt68xx.c instead.
Okay.
>
>
> In stop scan:
> > {
> > GT68xx_Packet req;
> >
> > + if (dev->model->flags & GT68XX_FLAG_SHEET_FED)
> > + {
> > + memset (req, 0, sizeof (req));
> > + req[0] = 0x42;
> > + req[1] = 0x01;
>
> Not applied. Used gt6801_stop_scan instead in the new command_set.
Okay - Thanks for the new version, I'll test that the next days.
Ciao,
Gerhard
