Is this really happening with cursors created with the CREATE CURSOR command, or just those created by a SQL command? If the latter, try adding NOFILTER to the end of your SQL commands.

>There has been a behavior change between VFP 6 and VFP 9 which handles cursors slightly differently, and I think is causing my problem. In VFP 6 cursor files were created in 8.3 format. That is no longer the case. I am seeing file names in 13.3 format, e.g. 00002m89000cz.tmp (or .cdx if the cursor gets indexed). Furthermore, I see from tracing this particular process that the first 8 characters of different cursor file names -- of which this process has six open at the point where it is acting weird -- are non-unique. So there are other cursors named 00002m89000p2.tmp, 00002m89000t2.tmp, etc.
>
>I think what is happening is VFP 9 is getting these files mixed up. I am seeing VERY odd behavior while debugging this. One time through it will say one cursor (alias) is not found. The next time maybe it works okay. Then it says a different cursor is not found. One time I flipped over to the VFP window and switched work areas to browse one of the cursors. After doing that, reselecting the previously selected work area, and going back to the debugger window, I quickly got an error saying a cursor that was not involved in any of this, one created much earlier, was not found.
>
>I may have to do what you did, using tables instead of cursors, but hope it doesn't come to that. This application has cursors up the ying-yang so it would be a lot of code changes and testing. Making matters worse, it runs most of the time with error handling turned off (see no evil, hear no evil, speak no evil....) so problems often don't show up until later -- the program just continues on its merry way before failing somewhere else.
>
>Do you happen to know if there is a way to make VFP 9 create cursor files in 8.3 format?
>
>Thanks to all who have been helping!
>
>Mike