Overdue context sensitivity in OCR (Was: Best way to scan 132 column fanfold mid-1970s text printout

On 24 Jun 2011 at 20:59, Dan Gahlinger wrote: I don't think I could ever forget how--I spent too much of my life working on FORTRAN compilers. I got the DEC PDP-10 source code for Adventure from a DEC field engineer about 1974-75 on a reel of 9-track tape (don't laugh--we were still heavily into 7-track). I was surprised to see that it was text packed into binary form--3 7-bit ASCII characters per 36-bit word (you'd have thought that if it was character data, it'd be one character per tape character); converted it to 6 bit display code and unraveled the DEC-10-specific FORTRAN constructs into their CDC 6000 versions (harder than you'd think, particularly with Hollerith constants); transformed the database to be readable by 6000-series FORTRAN I/O and I was off and running. You have no idea how much company computer time was wasted on that thing. The COMSOURCE admins used to conduct sweeps of employee permanent files and purge every copy they could find. So, if this is anything like that DEC 10 version, I can probably lend a hand... --Chuck