dlialyzer a bit too sceptic?

[Ulf Wiger (AL/EAB)]

Dialyzer (R11B-0) gave me the following warning on some 'rdbms' code:

{rdbms_groups,'-do_add_members/2-fun-0-',2}: Call to function
{rdbms_props,do_write_property,2} with signature ((_,{{'acl',atom()} |
{'typedef',_} | {'attr',_,_} | 'access_module' | 'acl' | 'indexes' |
'read_filter' | 'rec_type' | 'references' | 'write_filter',_}) -> 'ok')
will fail since the arguments are of type
('schema',{{'group',_,'membership'},_})!


But the function rdbms_props:do_write_property/2 looks like this:


do_write_property(Tab, Prop) ->
    mnesia_schema:do_write_table_property(Tab, Prop).


I.e. absolutely no restrictions on Prop. If one were to introduce one,
it would be something similar to that of
mnesia_schema:write_table_property/2 - namely that Prop is a tuple with
at least one element.


write_table_property(Tab, Prop) when tuple(Prop), size(Prop) >= 1 ->
    schema_transaction(fun() -> do_write_table_property(Tab, Prop) end);
write_table_property(Tab, Prop) ->
    {aborted, {bad_type, Tab, Prop}}.

do_write_table_property(Tab, Prop) ->
    TidTs = get_tid_ts_and_lock(schema, write),
    {_, _, Ts} = TidTs,
    Store = Ts#tidstore.store,
    case change_prop_in_existing_op(Tab, Prop, write_property, Store) of
	true ->
	    ...
	    %% we have merged the table prop into the create_table op
	    ok;
	false ->
	    ...,
	    insert_schema_ops(TidTs, make_write_table_properties(Tab,
[Prop]))
    end.


It's not at all clear to me why Dialyzer thinks the call would fail.

BR,
Ulf W