Ann W. Harrison wrote:

>Firebird 2 changes the semantics of inserts, updates, and
>deletes from naturally updatable views with triggers.
>
>Reasonable people may disagree about whether that was a good
>thing. However, that is the way that InterBase and Firebird
>have worked for nearly 10 years. An update to a simple view
>with a trigger changes the database just as if it were an
>update to a base table with a trigger. If you want to do
>something clever, you have to make the view non-updateable -
>joining the single table with RDB$DATABASE for example.
>
>
>

Your arguments are, sigh, unassailable. If there are programs that do
it that way, that is the way we must support.

I do like your idea of putting a clause on view definition to explicitly
block "natural" updates. Do you have syntax to propose?