[sane-devel] [sane-commit] CVS update of sane-backends/frontend (saned.c)

[Ilia Sotnikov]

On Thu, Mar 12, 2009 at 10:36 PM, Julien BLACHE <jb at jblache.org> wrote:
> Two points:
>  - we've never had any complaints about that
>  - saned is routinely used as part of LTSP and they never complained
>   either

What complaints you need? I also used the SANE daemon for long time,
and wrote patches rather than complaining. It's a free software, free
as in speech.

The errors inside a program should be divided into fatal and non-fatal
ones. If one couldn't get connecting system IP address while calling
getpeername() and you're using that name just for logging, generally
the error is not fatal. But when you use the name to do access
restriction procedures, it turns into fatal one. Just a brief example.

And now the deeper question: for what purpose do we need to resolve
local system address at all? To do rarely existing network connections
to SANE machine from itself? Only in that case the check (local ==
remote) will be true and the permission will be granted.
Am I missed something?

-- 
 Ilia Sotnikov