> This worked (discovery) at one point (I tried it very early in OLPC > days). > > But printer discovery is better done by some other mechanism than that > defined in IPP anyway, which is stupid broadcast. Even MDNS is less > evil (e.g. avahi). > > Some work is probably needed for scaling of printer discovery. > > Running a cups server is almost certainly not necessary, though its > footprint might not be too terrible if it is stripped of all the stupid > printer definitions (which are huge). I haven't tried that experiment.
In theory one of the supposed advantages of cups is that clients don't need all the drivers and it can just be installed on servers. Clients then print to the servers using PS (I think there's a minimum version requirement) and if there's a need for specific drivers the server then does the conversion. That would work nicely I would have thought. The dependencies in Fedora would need some cleanup, and the Fedora guys have done some pretty cool stuff in python (I think) for printer config etc so I assume that work could be reused/refactored as part of the job (but its dependency on foomatic would need to fixed :-) ). One other issue is the cups DNS-SD backend is written in perl. See RHBZ 465157 I filed a while ago about this issue. Peter _______________________________________________ Devel mailing list Devel@lists.laptop.org http://lists.laptop.org/listinfo/devel
