When using ODBC, is there a way to have the backend act as pessimistic when someone pull a record to edit it? I mean, with direct table access, we use RLOCK() in order to lock the record. For example, when we want to increment the next key using the primary keys table, we apply a RLOCK() for a short instance in order to avoid someone else to change that value while we are changing it. This is what I would like to obtain using ODBC. So far, in all the samples, it seems that we pull a set of data or one record to edit. Then, at save time, we will see if our save is successful. The requirement of one client presently is to use ODBC and apply a lock on the backend in order to make sure that a record is not modified by 2 persons.

What would be a bullet proof mecanism for this case?