Question

我有两条记录：

-record(foo, {timestamp, name}).
-record(bar, {timestamp, name}).

我想执行一个模仿以下SQL查询的Mnesia查询

SELECT f.* FROM foo f WHERE f.timestamp NOT IN ( SELECT b.timestamp FROM boo b)

什么是有效的Mnesia等价物？

Answer 1

好问题！现在，我想到了两种方式。我们使用qlc的地方和我们使用mnesia's own table iteration methods with accumulators的地方。这是第一个选择：

%% Here i use 'qlc', with a guard
%% which ensures that the timestamp
%% in the given 'foo record'
%% does NOT exist in table 'bar'

query()->
    Is_not_in_bar = fun(Stamp)-> 
                     case mnesia:dirty_read({bar,Stamp}) of
                        [] -> true;
                        _ -> false
                     end
                end,

    Q = qlc:q([R || R <- mnesia:table(foo),
                         Is_not_in_bar(R#foo.timestamp) == true])),
    Fun = fun(QH)-> qlc:e(QH) end,
    mnesia:activity(transaction,Fun,[Q],mnesia_frag).

另一种选择是在交叉引用表foo中的每个时间戳时迭代表bar。如果在bar中找不到它，则将其添加到累计金额中。看下面这个

%% Here i iterate table 'foo'
%% for each record i find,
%% i try to cross reference
%% its timestamp in table 'bar'
%% If its not found, then i add that
%% 'foo record' into the Buffer
%% I accumulate this Buffer until
%% the entire table 'foo' has been 
%% traversed

query_vsn2()->
    Iterator = fun(#foo{timestamp = T} = Rec,Buffer)->
                    NewBuffer = case mnesia:dirty_read({bar,T}) of
                                    [] -> [Rec|Buffer];
                                    _ -> Buffer
                                end,
                    NewBuffer
                end,
    Find = fun(Loop)-> mnesia:foldl(Loop,[],foo) end,
    mnesia:activity(transaction,Find,[Iterator],mnesia_frag).

我想根据表大小，应用程序和用户偏好，这些功能中的每一个都会产生影响。但是，试试它们，看看哪一个很好地融入你的应用程序。好消息是，这完全是一个读取工作，没有写入，所以我期望足够高效。成功！

Mnesia相当于SQL NOT IN

1 个答案: