This was very helpfull.
thanks
This was very helpfull.
thanks
You'll need to install either the samsungmfp-configurator-qt3 or -qt4 package as well. (Believe me - I'd love to simplify the complexity of the number of packages.) Good luck with the router/USB hookup, but I think it should work at this point. I'm not sure why the different port generated the usb errors; I used to see reports of issues like that with hubs all the time and had some bad experiences myself a long time ago, but less commonly in recent years.
You've explained as much as I really understand. However, yahs' post suggests that the issue has been resolved (at least for now) with Natty, so there may not be much point. I doubt the Ubuntu devs will release a fix for Lucid at this point, even though it is LTS. It's still a valid bug, but if you can deal with it by upgrading to 10.10 and the problem does reappear, it's not likely to get much attention.
The first post is a regularly updated summary of (nearly) everything important in all the other comments. Basically, my recommend if the printer doesn't "just work" is to install the Samsung drivers via my repository as described in the first post. All the caveats and whatnot are there only if you run into problems.
There's one extra problem with the Samsung drivers: the scanner can only be used within the same network segment, it can't be reached from a machine behind a router or firewall. (Printing works fine, just enter the machine name/address manually.)
Here's a workaround for those in this situation (works for me):
replace /opt/Samsung/mfp/bin/netdiscovery with this:
#!/bin/sh
if [ -f $HOME/.samsung.netdiscovery ]
then cat $HOME/.samsung.netdiscovery
else
LD_LIBRARY_PATH=/opt/Samsung/eglibc32/lib /opt/Samsung/mfp/bin/netdiscovery32 $*
fi
Then, in some machine that *is* in the same network segment with the Samsung, do
/opt/Samsung/mfp/bin/netdiscovery >.samsung.netdiscovery
and copy the resulting .samsung.netdiscovery to the machine(s) in other network segments.
If you have several Samsung's in different network segments this obviously won't work, but I think simply appending netdiscovery results from each might do it (I don't have many of them son can't test this).
The main reason why I built a 32bit glibc to work around this problem was because I knew it wouldn't conflict with any of the native utilities on my system (i.e. the shell in this case). It seems silly, but would it be possible to build a 64bit glibc for 32 bit systems and use the 64 bit version of netdiscovery?
I think another way to do it would be to build a shell against the new glibc and distribute the shell binary as well.
No. 32-bit libraries will work on a 64-bit system, but not the other way around. I agree it would probably solve the problem if it would work, but the actual implementation requires completely isolating the 64-bit processes and is a lot more involved.
Actually, you probably could just compile a new shell, even without worrying about the new glibc. However, I haven't figured out how to package such a solution. I can't actually distribute a recompiled shell for general use, because installation would conflict too severely to be resolved. And I haven't worked out a solution that would reliably isolate all the Samsung-related process to their own shell (or chroot, or other separate process area) and still be general enough to package up. (If I ever have a few solid days to work on this, which seems unlikely, I might get it to work, in which case most of these other ongoing problems could also be dealt with. But that's perhaps just wishful thinking and not realistic given my free time right now.)
Bookmarks