>I have a scenario where I am processing a batch of changes to a table. The
>changes can result in an Insert or a Delete. So I Fill a DataSet with the
>table's current contents, then I process the changes, then call Update on
>my SqlAdapter. Psuedo Code:
>
> Fill(DataSet, MyTable)
> Foreach( Statement in UpdateCollection)
> {
> Foreach(Row in DataSet.Tables[MyTable].Rows)
> {
> if(Statement matches Row & !found)
> Row.Delete()
> found=true
> }
> if(!found)
> DataSet.Tables[0].Rows.Add(NewRowFromStatement)
> }
> Update(DataSet, MyTable)
>
> Unfortunately, there is a small chance that two records in the current
> batch of changes are related, basically one record will be the delete for
> one of the inserts in the update collection and this seems to cause a host
> of problems I cannot get arround. If I call Update, then since the Deleted
> row never had an original version the DeleteCommand can't find an original
> row to delete from the source table and throws an exception (Deleted 0
> rows, expected 1). I had thought maybe if I found a row that was state
> Added and I needed to Delete it I could just call Remove, or
> AcceptChanges->Delete->Acceptchanges to cause the row to essentially
> disappear from the scope of the update. But this blows up the enumeration
> since it alters the Rows collection. I think my only option is to add a
> new column to my table MarkedForDelete and change the code to only Add and
> Update then make a seperate process to cull rows marked for delete
> asynchronously.
>
> Any tips?
>
> Thanks,
> Jason
>