Hi Michel,

a few tears ago I discovered that PhDbase on a big DBF is not reliable, and may even cause VFP to crash. The fact that PhDbase now is discontinued may indicate that this problem simply is not fixable.

In my case the DBF had about 1.5 million records, and among many things I found that PhDbase did not include all the records in its index file, even if I specified a complete index. The records which were missing in the index, did not include any 'common' words, and I found that for instance words starting with 'NJU' were NEVER included in the index. Since I sell integrated circuits whose name started with NJU, this was a HUGE problem, of course.

I spent very much time trying to get PhDbase to work, one of the solutions I tried ment splitting my bigger tables into smaller tables, perform searches on all the smaller tables and merging the results into one result cursor. It was possible, but far from practical.

So I made my own functions and procedures which give more or less the same functionality as PhDbase. In my system I did not need any wildcard searching, so if you need that, my solution will need a lot of more work. If you are interested in hearing more about how I fixed it, please send me a private email and I will give you the details.

>This is the first time I tried to index PHDBase on a one million record tables. The command I use is the same as usual:
>
>

>PHD('INDEX NOTES,TITLE,KEEP ALL,NUMBERS,FULL,LARGE')
>

>
>However, after a while, I get:
>
>Not enough memory for file map.
>
>I have 1 GIG RAM on that server. So, what is the problem?