[sane-devel] gamma table value

[Major A]

Fred,

I think I've understood some of the problems you're having with
GNU/Linux and SANE, so let me comment on your email.

> The problem with most of this material (like, in my experience, MOST
> computer articles, guides, HOWTOs and so-called 'Help' files), is that they
> are written by geeks for geeks, and assume that you know how to do things
> before you start. This leaves a humble, novice, never-been-there-before

There is a hierarchy of documentation for GNU/Linux software
available, ranging from introductory books meant for the novice to
make them acquinted with the ideas behind UNIX and Linux, through user
and administrator manuals down to programming manuals and
specifications for those who know everything else and want to write
some useful code that interfaces with what others have done.

Trying to use GNU/Linux for the first time and trying to read anything
else than the introductory books is like trying to drive a coach the
first time you see one. It's a much better idea to travel in one first
as a passenger to get an idea of how a coach works.

> USER, not only cold, but out in the cold. Quite often, I'm not able to get
> beyond the first paragraph before my eyes glaze over and I throw the
> offending paper down in disgust with an expletive or two, exclaiming (as our
> American friends say: 'Excuse me!? Come again!?').

You quite clearly tried to read the wrong document then.

> In going through the documents I started with, I reluctantly came to the
> conclusion that the only one which was suitable for my level of
> understanding was the Sheriff (even tho it was somewhat dated), because he

This is the one BIG problem with all computer-related
documentation. Whereas you can read about driving a coach from books
that are 20 years old, a document that is a few months old might very
well be outdated in the computer world. You must make sure you always
have up-to-date material, and as soon as you come across something you
cannot find anywhere on the internet, you must get suspicious that
what you're reading might not be the latest.

> obviously understood my predicament precisely, and wrote accordingly - with
> every command clearly set out. The rest were, to a varying degree, uneven

Please understand that this doesn't work, it's never going to. The
entire idea behind today's software is that everyone has total freedom
to install software on their computers as they wish. Therefore, typing
in commands from a book verbatim will sooner or later give you an
error message because something isn't installed or is in a different
place from what the book expects.

Back to my analogy: doing this is precisely like driving a bus using a
special atlas that tells you when (not where!) to push the brake or
turn the steering wheel by x degrees.

The only way around this is to learn the basics so you can either read
the more advanced and more specific literature or be confident enough
to work out from those 1:1 commands in Sheriff (whatever that might
be) what you have to do.

> So I went with Sheriff. He starts by telling me to remove glib, gtk+, sane
> and xsane - 'in order to install new and up-to-the-minute libraries to make
> SURE that all compiles.' This I do, using YaST2. Then he gets me to copy
> ($cp -a /usr/src/linux/include/scsi /usr/include). This is not a good start

See? You haven't got the kernel source installed, probably because
you've never built a kernel. Did Sheriff tell you to? If so, why isn't
there a distribution with all the Sheriff instructions already carried
out? (Besides, copying those files is the wrong thing to do anyway.)

> for me as I get the response: 'No such file or directory.' However, I press
> on and, as instructed, create a new directory (/home/<username>/scanner),
> into which I download and install glib-2.2.1.tar.gz, gtk+-2.2.1.tar.gz,
> sane-backends-1.0.11, and xsane-0.90, as described in my original message.

So, according to Sheriff, you tried to build all these from source
code. This used to be necessary in the old days (up until about 5
years ago), but these days, most distributions will give you very good
facilities for installing pre-built packages. That way, you can build
sane-backends and xsane from source without rebuilding almost
everything else that is required by these two.

> > I guess it would have been even easier to just use a SuSE binary RPM,
> > e.g. this inofficial one from
> > http://packman.links2linux.de/index.php4?action=211 (not tested by me).
>
> Excuse me, Henning!? Come again!? I'm sorry, but you've lost me already:
> I've no idea what a 'SuSE binary RPM' is!

You're trying to make a scanner work with GNU/Linux, and as such, you
are a system administrator. Therefore you must know -- at least --
what distribution you use (SuSE in your case), and from there can work
out from information on the internet how the distribution installs and
manages packages on your computer. (Besides, I'm sure every
introductory book mentiones RPMs these days.)

> It added: 'you may need to remove /dev/null before you can run configure
> again.'

(Aehhh, anyone know where this could come from? It's not supposed to
refer to /dev/null, is it?)

> > The problem is: You just can't make a "good, simple, clear
> > documentation that supposes no previous knowledge". That would be a
> > big book if it should tell details for every supported scanner.
> 
> Then that's a major shortcoming of Linux!

No, it's not: as I said above, it's all there, it's just split up in a
hierarchy. Windows, MacOSX, etc. are all the same -- you start with
those "Windows for dummies" books and finally end up at something more
advanced that certainly requires some background knowledge to
read. (Except that with Windows, many of the more advanced documents
such as programming specifications are simply missing.)

> > A short, simple instruction is: Install sane-backends, run scanimage
> > -L. If it doesn't work, read man sane, man sane-backendname.
> 
>  But I did install sane-backends and ran scanimage -L and
> sane-find-scanner - with the results detailed below - which, you say: 'Looks
> fine',  So I had no reason to think that there is anything wrong with
> sane-backends - if there is(?).

Why did you remove it then? Both sane-frontends and xsane build on it,
sane-backends is totally independent of these frontends.

> > > but, when I try
> > > to launch scanner [as 'root'] by clicking desktop 'Scanner' icon, I get:
> > > 'KDElnit could not launch "xscanimage": Could not find "xscanimage"
> > > executable'.
> >
> > Neither sane-frontends nor sane-backends contains a "desktop Scanner
> >icon".
> 
> On my SuSE desktop - both 'root' and 'user' versions, which are slightly
> different - there are a number of icons, such as Trash, /Windows/C,
> /Windows/D,
> OpenOffice, CD-R, CD-ROM, etc, etc, including one marked: 'Scanner'.
> This is the icon I clicked.

This is nothing to do with SANE. You may think otherwise, but the
command line is a very basic instrument, not only in GNU/Linux but
also in Windows (in fact, more than before with its modern
super-intelligent versions).

I can only guess (don't really know SuSE nor KDE or whatever desktop
you use) that the "scanner" button simply tries to run xscanimage,
which is equivalent to you typing "xscanimage" on the command line.

> As I stated in my original message, I have also downloaded/installed xsane
> as it seems to be generally recommended., but haven't yet tried to use it
> (none of the articles I have cited tells me how!)

Well, if you could build scanimage and run it, then you'll certainly
be able to do this one as well -- the basic procedure is exactly the
same.

Hope I didn't upset anyone, didn't mean to.

  Andras

===========================================================================
Major Andras
    e-mail: andras at users.sourceforge.net
    www:    http://andras.webhop.org/
===========================================================================