>>I guess this is just a VISUAL Fox forum but I know a lot of you all go way back with FoxPro. This is the first time I've seen the Universal Thread - Sweet! I'm doing other stuff these days but get called in to support an old dbase app that is being run in FoxPro for DOS 2.6 (b?). The latest problem is a real stumper!
>>
>>Records are being misteriously marked for deletion when they shouldn't be. EVEN MORE MISTERIOUS - THESE FILES WON'T PACK; NO ERROR - THEY JUST DON'T GO AWAY.
>>
>>Has anyone seen anything like this. I know its not much to go on but I'm at a loss to even know where to start. Very little code has been changed since April '96 and the behavior just started in October '96 (or later). We can't get it to reproduce either but it has happened three times now. Questions have been raised about operating system interference, index corruption, file corruption - you name it. I was leaning towards programming error but the fact that the records won't pack have me scratching my head. They've been running the app under windows - I suggested they stick to DOS for awhile.
>>
>>Any suggestions would be greatly appreciated. Thanks!
>
>carol, have you by any chance added a lan in addition to windows? you may want to try adding 'exclusive' to your USE statement (may as well start getting use to thinking multiuser anyway). without more information, this is my best guess.

Thanks, I'll look into it but they've been on a lan all along. Because they are sharing programs only and not data we have set exclusive on already. What other kind of information would you like.

We are using a foundation read to run programs written in dbase through a simple menu like screen. They have ndx (or is it idx - I get those two confused) indexes, not cdx. Data is keypunched into an ASCII program, then ran though a fortran program to create multiple text files which are loaded into FoxPro which then rearranges them into other dbf's (These programs were written by forresters and arn't real pretty). Then the dbf's are ran though SEVERAL programs to validate the data. Finally the data is 'archived' (deleted) and offloaded to a UNIX system.

They first suspected the keypunch data but can't find anything wrong with it and like I said - when they reload the data the error doesn't reproduce. The records have a two field key with the second field being many to one of the first key. One clue is that it is always the first record in a set that is being corrupted (that's why I suspected a programming error). Something that I didn't add is that the first key is being corrupted and receiving control characters some how. I thought I'd just see if anyone's ever experienced the won't pack phenomenon before just as a clue to all the other problems. Without being able to recreate the error Fox's debugger won't help much. (An excellent debugger I might add - I miss this debugger!)