Pit M. wrote:
>
>> It would have failed if you had run the cursor far enough to fetch one
>> of the bad rows.
>>
>> regards, tom lane
>>
> The difference is that in one case the query fails and in the other the
> FETCH command fails.
>
>
> Our problem is that if a query succeeds we use a count(*) of that query
> to display the result count in the status bar of our application - the
> select works but the count(*) fails :-)
> When we use this query on MS SQL-Server the query fails directly though
> we also use cursors here. So the SQL-Server somehow checks all the data
> only for the query -> perhaps because we used a server side cursor.
IIRC, the behavior of MSSQL will depend on the query plan. If it's a
plan that requires doesn't require materialization at all, it won't
figure it out until you get there.
//Magnus