Greg,

I have read David's reply to your question.
SNIP
>
>With an 40+GB hard drive on most of my systems, is cdx bloat something I really need to worry about? :) I don';t care how big they are as long as they work like they're supposed too.

The file size limit of any kind in a VFP system is 2 gigabytes per file. That includes .DBFs and their variants and .CDXs and virtually any other file type used in VFP.
David says that he would be inclined to move to a server if a system was approaching that size with any of its tables (files).
While that is perfectly legitimate, I just want to point out that many of us have/do (me not any more, in my current situation) worked with application systems that had many files near the 2gig limit and had perfectly acceptable performance with VFP and file-server technology (not peer-to-peer). Yes, it involves more care and extra work, but there are trade-offs going to a server environment too.

In the side-issue at hand there is every indication that these larger .CDX files do not at all impact the reliability/accuracy of VFP data or its processing (as I read Christof's response). There may be some penalty in terms of performance compared to the prior method used and there certainly is a penalty paid in terms of space efficiency. If it turns out that the changes adopted were in response to fixing up some data-integrity bugs that may have existed in the prior version(s) of VFP, then I wouldn't be inclined to call them "penalties", but rather a required overhead. That, I guess, remains to be seen.

The issue of THIS thread is that REINDEX now gives a different result using the same set of data. If this is as a result of NECESSARY changes (described earlier) then that will be fine. Until that is confirmed as the answer, I consider it to be a bug.

I think that it was VFP6 SP3 that introduced a much faster REINDEX performance. The VFP Team said nothing about it when SP3 was delivered but it was noticed and confirmed shortly after people implemented SP3.
While I think the VFP Team would have done themselves a favour by noting the improvement in the SP3 notes, there clearly was no potential harm in doing otherwise.
The two issues at hand, on the other hand, can have an impact on many deployed applications so I think it was worthy of mention in the VFP7 release notes. In fact it seems reasonable to surmise, in the absence of such notice, that the results obtained are unintended.

Jim