[sane-devel] saned, Network Scanner, Automatic Document Feeder

[Nicolas]

Good news then, I've included this fix for next pixma CVS commit.

BTW, my answers to your questions in the text below, 
I copy also the Sane-dev mailing list too, as those points regarding ADF
are interesting:

Le samedi 04 octobre 2008 à 00:55 +0200, Pascal Seiler a écrit :
> Hey Nicolas, your great!!!
> 
> I did a few tests and all worked perfect. At the beginning of testing I 
> had a situation where the scanner hung and I had to power cycle it. I 
> can't reproduce that, but I think it was murphy. :-) Here's what I did 
> so far (all over the net in debug mode (SANE_DEBUG_PIXMA=10, scaned 
> -d128, as well as with inetd):
> 
> scanimage flatbed from command line with and without batch-count
> scanimage adf from command line with and without batch-count
> 
> gscan2pdf in scanimage mode as well as with scanadf, both with page 
> count and with continuous mode (all pages)
> 
> The log files in flatbed mode are produced with 'batch-count' (otherwise 
> the log would get endless) ;-) and in ADF mode I did without.
> 
> flatbed.log
> scanimage --mode=Color --source='flatbed' -y 2 -l 0 --resolution=300 -x 
> 2 -t 0 --batch --progress --batch-count=3 > tmp/scantest -v
> 
> adf.log
> scanimage --mode=Color --source='Automatic Document Feeder' -y 2 -l 0 
> --resolution=300 -x 2 -t 0 --batch --progress > tmp/scantest -v
> 
> Here are a few more questions:
> - Am I right that I need the new files only on the backend?
> - For other clients can I use the packages that ships with the distribution?
Yes, the patch file I sent you is for the backend, running on your
server.
For the other clients which uses the frontend, this file is not necessary, 
you can run the frontends coming with your distribution.

> 
> - I think this is related to the graphical frontends, but I mayask 
> anyway. When scanning with xsane and gscan2pdf I get this error messages 
> when the ADF gets out of documents.
> 
> 	- xsane
> 	 error during read: Document feeder out of documents.
> 	
> 	- gscan2pdf
> 	Unknown message: scanimage: sane_read: Document feeder out of documents
Yes, this is working as designed, those messages are returned by the
frontend because the backend informs it about "No Paper", when its ADF
is empty.

> 
> For sure this is a cosmetical error and if canceled I can finish or 
> continue the job but it would be nice to have more convenient messages.
> 
> For example:
> If a number of pages to scan is defined and ADF is empty, give a message 
>   with the option to either load ADF and continue toscan or to cancel 
> the scan job.
> If the 'all page' option is selected the job should just silently finish.
IMHO, this is related to frontend only, but maybe I'm wrong ?

In a more general way, I think it would be fine if the backend could
tell the frontend, at the end of a page scan: "My ADF is not empty,
could you request me another page to scan" ? 
This way, it wouldn't be necessary to specify the # of pages to scan,
this would act as a copier, feeding paper automatically as long as
there's paper in the ADF.
Is that possible with Sane ?

> 
> Anyway, many many thanks for your great efforts. I really appreciate 
> your work. If you need more testing and/or other logs, please don't 
> hesitate to contact me.
> 
> I'll do as much I can (with my limited knowledge) to help you.
> 
> Best Regards and have a nice weekend
> Pascal
> 
> *******************************************************************************
> "The only intuitive interface is the nipple. Everything else is learned."
> 
> 
> Nicolas wrote:
> > Sorry for that!
> > 
> > After copying the 3 files, simply do the "make" and "make install"
> > steps. 
> > You should see a recompilation of the pixma*c files.
> > 
> > Nicolas
> > 
> > Le vendredi 03 octobre 2008 à 23:13 +0200, Pascal Seiler a écrit :
> >> Nicolas,
> >>
> >> I'm far from an expert ;-) what exactly do I need to do? ./configure, 
> >> make, make install or make, make install, or just a make install?
> >>
> >> Pascal
> >>
> >> Nicolas wrote:
> >>> Before committing the current release, could you give a try with the 3
> >>> modified files attached
> >>>
> >>> Place them over existing CVS ones in directory sane-backends/backend and
> >>> rebuild/reinstall Sane
> >>>
> >>> Let me know what logs you will get in flatbed and ADF
> >>>
> >>> Nicolas
> >>>
> >>> Le vendredi 03 octobre 2008 à 21:47 +0200, Pascal Seiler a écrit :
> >>>> Ok, thank you very much. I'll be waiting for it.
> >>>>
> >>>> Pascal
> >>>>
> >>>> Nicolas wrote:
> >>>>> Have some presumptions on where this could come from, as ADF code in the
> >>>>> pixma backend isn't very "clear". 
> >>>>> But before any change, I'd like to submit a newer version of the
> >>>>> backend, that should enhance ADF scanning (tested on MX850 currently).
> >>>>> I'll keep you posted when commit to CVS, as I've some other fresh
> >>>>> ongoing changes for MP970 TPU too, and BJNP network support, thanks to
> >>>>> Louis Lagendijk.
> >>>>>
> >>>>> Nicolas
> >>>>>
> >>>>> Le vendredi 03 octobre 2008 à 09:17 +0200, Pascal Seiler a écrit :
> >>>>>> BTW,
> >>>>>>
> >>>>>> this is how the log looks like when I to the same with source=flatbed.
> >>>>>> Maybe that helps.
> >>>>>>
> >>>>>> Pascal
> >>>>>>
> >>>>>> Nicolas wrote:
> >>>>>>> Yes, this is exactly what is needed, but looks like there's only a
> >>>>>>> single page scan here.
> >>>>>>>
> >>>>>>> Could you produce a same trace (but select a _small_ scan area for
> >>>>>>> smaller log size) of a 2 page ADF scan with the issue you get.
> >>>>>>>
> >>>>>>> Note that I'll commit very soon some changes that have been tweaked with
> >>>>>>> a MX850 (but should suit MX310 as well) that enhances speed for ADF
> >>>>>>> scanning, and imitate exactly what the Doze driver does. I'm just
> >>>>>>> waiting to have also MP970 TPU scan almost ready.
> >>>>>>>
> >>>>>>> Nicolas
> >>>>>>>
> >>>>>>> Le vendredi 03 octobre 2008 à 02:03 +0200, Pascal Seiler a écrit :
> >>>>>>>> Not sure about the difference, but now it has worked.
> >>>>>>>>
> >>>>>>>> See the attached file for the output
> >>>>>>>>
> >>>>>>>> Pascal