The old behavior seems ok to me. The current info at least reflected
what the actual stats were at some time, which seems more likely to
reflect the future stats than an out of blue estimate by the optimizer
if the stats are deleted.
/mikem
Kathey Marsden wrote:
This is kind of an edge case, but I wonder, what should be the behavior
of update statistics on an empty table if it had previously created
statistics.
The current behavior seems to be that it leaves the old statistics
values in sys.sysstatistics. Is that correct? It seems to me it should
drop the statistics information in this case.
Thanks
Kathey
