>>No. The RI code will not be touched.
>>
>>>
>>>ok, so i'll drop all tags and recreate all indexes with the help of a table containing all inforation abouts dbcs and tables. But what about the RI code? I'm using the RI code generated by VFP. Do I need to recreate RI code after a complete recreation of the indexes or I can eep using the old one (figuring that i won't change the relations in the DBCs)?
>>>
>>>Thanks one more time for your help,
>>>
>>>Alonso.
>
>OK, great!
>
>THanks, Craig & Tenea!

Hi Alonso,

I just made a 100,000 record table with 4 biggish indexes. Once created it was 3,419KB and its CDX was 2,859KB.

I then closed it, opened exclusive, and did a REINDEX. Its CDX size decreased to 2,015KB. Ignoring "bloat" for the moment, I think this makes sense because I purposely made 2 of the indexes content exactly backward (descended in value).

I then closed the table, opened it, and did 5 REINDEX commands in a row.

When I closed it the CDX was (still) 2,015KB.

I think it is fair to conclude that the REINDEX command, at least as of SP3, NO LONGER CAUSES CDX BLOAT.

So REINDEX just might be a simpler solution for you all around *IF* you are not doing this specifically to recover from a corruption problem.

Cheers,

JimN