>I'd only suggested using table and memo fields because he'd stated he was having trouble with processing the data otherwise. If it was me, I'd likely approach it with low-level file routines and essentially process the file on a line-by-line basis. My initial impulse would to implement the program as an AWK script.

Perhaps nowadays the machines are fast enough to make no difference, but once upon a time I gained most speed using a combination of what Craig said (fread() using decent sized chunks - 16K then, perhaps 64K now), then processing each chunk row by row until there was less than one row left. Then append the next chunk to that partial road, process that, while not feof() (first I wrote until feof(), but we don't have until in fox).