I've had this problem off and on for some time, but now it is getting ridiculous. I search through this discussion group to find a similar problem, but couldn't find one.

I build an index (this happens with both BTree and CrossRef)and everything appears fine until I do a search. If I have one record that should be found, AREV comes up with 2 records and these 2 records are not proper records. For example, I search for 12345 in a file, e.g. SELECT FILE WITH FIELD=12345'

The record containing the 12345 is ABCDE, but the search comes up with to records, not one. The search comes up with ABC and E, with two parts of a record, not the whole record.

There is no ABC or E in the file, only ABCDE and only that record contains 12345. If I remove the index, then the search will select only one record, i.e. record ABCDE.

Any suggestions. BTW, the file is not Quickdex'ed.

Frank,

My suggestion is to apply all the normal safety checks with your indexed fields. A couple of initial items to check are:

1. The dictionary length is defined to be at least one to two characters longer than the longest data being stored here

2. There are no delimeters in any key ID's

dbakke@srpcs.com

SRP Computer Solutions

In fact, I'd be willing to bet that there is a system delimiter in the key field. It's usually a value-mark but a field-mark can be equally troublesome to get to. Also, I've seen a binary zero get into a key like ABC(char(7))DEF. Won't show with R/LIST but will get processed during index creation.

Don Miller

C3 Inc.

Thanks Don and Don. I'll try your suggestions. Sorry – I didn't get back to this forum since I was busy with my programming.

View this thread on the forum...