[sane-devel] segfault in sane_net_get_devicesconnecting to network saned

[Brian J. Murrell]

Julien BLACHE <jb <at> jblache.org> writes:
> 
> Hi,

Hi,

> That's what you get for using a proprietary backend that isn't properly
> integrated in the system (udev/consolekit).

So, is this just comment on the permissions issue or the the segfault -- which
is the original point of this report/thread?
 
> Though Ubuntu could do a better job when it comes to SANE, they're not
> really at fault for this one.

Well, there is something, which I have yet to go dig up and apply that fixes the 
issue, so yes, there is something more that Ubuntu can do.

And you don't need to preach to me about binary drivers.  I know all about it 
and in fact did a round or two with somebody on the open-printing list about 
warning others about the evilness of the support for this printer.  To my 
chagrin they were entirely forgiving about it, being thankful that Brother 
supplied at least binary drivers.  :-(

But the bottom line is that this printer was a gift from a well-meaning Mom who 
doesn't know (and who I wouldn't expect to know) about the politics of hardware, 
software and open standards, so I can either have it doing nothing more than 
holding down paper or I can try to work within the boundaries set by my well-
meaning Mom.

Ultimately, xsane works with the binary driver and saned doesn't.  To me 
(admittedly, an outsider) at least, it sounds like saned is doing and/or
expecting something that xsane isn't and if we could determine what that 
difference in handling is, we could fix a segfault in saned.

Put another way, perhaps the segfault is in the binary driver.  That will be 
unfortunate if it is.  But perhaps it's in saned/libsane, in which case there is
a bug in saned/libsane (a segfault is undeniably a bug).  But even if the 
segfault is in the binary driver, we can't deny that the binary driver works 
with xsane, so saned/libsane might be able to avoid tickling whatever bug is in 
it.

Thots?

b.