john@minster.york.ac.uk (06/13/91)
This may be a frequently-asked-question. :-) Does your company/department use a desk top scanner to capture text and graphics for use on a UNIX system? If so, how is the system arranged? What text-recognition software did you use? (Ideally, I'd like public-domain C source.) I've found several bits of hardware (SCSI-interface preferred), so that isn't my problem. What I need to know is how best to make it available to the naivest of users. Of course, my budget isn't too large. :-) I can connect the scanner to a PC running one of the MS-DOS text-recognition packages, and that to our TCP/IP LAN with something like PC TCP. More straightforwardly, I can plug it into the back of a Sun 3/50 and read/write the /dev/s[dt]0 SCSI device. This last mode, however, probably means dedicating the Sun to this use - which seems wasteful. Also, there is then probably the need to buy expensive, quirky, binary-only software. My ideal would be a scanner with a dedicated ethernet interface and TCP/IP, a one-line 80 column LCD display and a plug for a cheap keyboard. It would prompt `login:', then `password:' and finally `pathname (control-D to logout):'. You'd place your document on the glass, type the file's pathname and press return. It would prompt for another file. Whether I'd want this program also to do my image-to-text conversion I don't know. Perhaps this would be best done with a filter once the person was back sitting at their own workstation. The systems administrator here recommends omitting the password prompt and avoiding the need for the box to do authentication - the scanned images would end up owned by `user' in /usr/spool/scan/`user'/`name', where `user' is the login name, and `name' is the simple file name given. Yes, I have chased up leads from Sun's third-party hardware/software catalogue "Catalyst". :-) It's recommendations I'm after. John A. Murdie