What you say confirms my fears. Is there any way around this without going to Asyncronous mode?

The same thing happens in VFP7 with SPT so the Cursor Adapters by themselves are not wrong.

Is there anywhere where I can get more info about this darn fetchsize?
Is it a Visual FoxPro problem or an ODBC problem?

This most certainly looks like a BUG, most noticable when the SQLServer dishes out the data for the first time. After that it ic cached and a bit faster.

Any ideas?


Sorry Bernard.

VFP ODBC have other problems,
but this is not a problem.

Yes, you can say this a bug because the Talk not show your FetchSize values,
but it is insignificant because:

With Syncronous mode you have a atomic process,
and this fetch MaxRecords rows.
If you set FetchSize = -1,1,100,1000,1000000 the process and speed not change.


On help is writted:
... and the remaining records will be fetched in blocks of 100 ...
this can be interpreted badly, because a block is a C cycle like this

-- foxpro equivalent metacode
for k=1 to fetchSize
  ....
  for column=1 to columnCount
     ODBCFetchColum
     ....
  next
next

Block is relative to the syncronous slice into the asyncronous process.

Then, ignore this false problem.

Fabio