>Hi!
>
>I specially agree with 1) and that is what I meant - provide a TP program created that .DAT file.
>
>3) might be not enough. Analizing of data require also to know the logical purpose of each data field, that is not such an easy thing.
>
Exactly. This reminds me of the thing I forgot to add:
4) Try to outweigh the amount of work needed for conversion versus the amount of work needed for data entry. In some cases, the second is easier. Also, using the old software to do some reporting and redirecting the reports into text files may be easier - at least these are ASCII files with some column descriptions and the format is more or less fixed. You get something that can be parsed.
I've had a case when I sweated for three days converting a loans' history table from big-iron IBM, where the floats were converted from EBCDIC into ASCII, so I had to write a conversion back to EBCDIC in order to get the original numbers, then to unpack them etc etc... in the end I was only half successful, some columns were right and some weren't. Then it turned out that out of 2000 loans (with about 18000 total records) only about 80 were still active, and they didn't need the history, only the present debt. One girl entered these 80 in two hours.