Greg Troxel wrote: > > I can certainly accept that I did things wrong. But given that I'm a > long time sane user, I'd say this is an indication that things are too > hard. > > It is pretty counterintuitive to me that device on the command line is > not simply a pathname to a device special file.
Yes, it is a bit counterintuitive if you compare it with other device names. But the format of an xscanimage device name is mentioned in the man page. > > Once I put in the /dev/scanner link, xscanimage worked fine. Clearly > then, I did not have to tell xscanimage that my scanner was a umax. You can define device file names in most backend config files, and most if not all SCSI backends try to open /dev/scanner. > So perhaps there should be an OS-dependent notion of a list of default > files to check for (SCSI and usb scanners, or pass/ugen things that > probe as scanners). Well, there is something like that: the function sanei_scsi_find_devices. But it seems that nobody bothered yet to write an implementation of this function for NetBSD. > Basically, I think it comes down to that if sane-find-scanners can find > the scanner, then xscanimage and scanimage should find them as well > when started with no arguments. > > (Or perhaps sane-find-scanners should record them in ~/.sane or > something.) > > I'm not trying to complain - I'm posting in the hope that small > changes (that I don't know how to make since I haven't looked at the > code in several years) might make it easier for future users. You're welcome to write sanei_scsi_find_devices for NetBSD ;) Abel
