Erlang Thursday – Using ETS select with a limit

In last week’s Erlang Thursday we continued to explore ets:select/2 and seeing its use when combined with using ets:fun2ms to generate the match_spec()s.

This week we will take a look at the other versions of select that the ets module provides.

Yet again we will do our new playground ETS table setup, so if we crash our shell session we don’t loose the table.

Fun = fun() -> receive after infinity -> ok end end.
% #Fun<erl_eval.20.54118792>
SomeProcess = spawn(Fun).
% <0.52.0>
TestTable = ets:new(ets_table, [public]).
% 16402
ets:give_away(TestTable, SomeProcess, []).
% true

Next we will load our test ETS table with a bunch of test “products”. For ease of example, we will just use a number for the product id, and a random price ending in .99.

[[ets:insert(TestTable, {ProductId, random:uniform(100) + 0.99})
  || ProductId <- lists:seq(1, 10000) ]].
% [[true,true,true,true,true,true,true,true,true,true,true,
%   true,true,true,true,true,true,true,true,true,true,true,true,
%   true,true,true,true,true|...]]

We will create a match_spec() to find items in their twenties (and we will go ahead and round 19.99 up to 20 just because).

ProductsInTheTwenties = ets:fun2ms(fun({Product, Price})
                                     when Price >= 19.99 andalso Price < 30
                                     -> {Product, Price}
                                   end).
% [{{'$1','$2'},
%   [{'andalso',{'>=','$2',19.99},{'<','$2',30}}],
%   [{{'$1','$2'}}]}]

And if we use ets:select/2 against our table with this match spec, we get all of the results back in one query as we saw previously.

ets:select(TestTable, ProductsInTheTwenties).
% [{4351,29.99},
%  {635,19.99},
%  {6005,20.99},
%  {3742,27.99},
%  {5956,29.99},
%  {3753,28.99},
%  {6653,25.99},
%  {5151,28.99},
%  {2693,27.99},
%  {4253,21.99},
%  {7636,23.99},
%  {1935,19.99},
%  {9044,22.99},
%  {7797,22.99},
%  {2147,23.99},
%  {2574,26.99},
%  {7575,29.99},
%  {2130,28.99},
%  {4908,27.99},
%  {2218,22.99},
%  {9848,21.99},
%  {7632,26.99},
%  {3562,21.99},
%  {3130,27.99},
%  {575,26.99},
%  {4622,28.99},
%  {5678,25.99},
%  {4022,...},
%  {...}|...]

But the ets module also gives us a way to limit the results if we would like, using ets:select/3 and giving a limit of the number of results to return at a time.

So let’s use ets:select/3 and give a limit of 10 and see what happens.

ets:select(TestTable, ProductsInTheTwenties, 10).
% {[{9027,27.99},
%   {7347,29.99},
%   {7282,20.99},
%   {9386,24.99},
%   {5415,25.99},
%   {4032,29.99},
%   {8105,25.99},
%   {4634,24.99},
%   {1275,20.99},
%   {234,20.99}],
%  {16402,576,10,<<>>,[],0}}

We get a tuple back instead of a list of results. The first item in the tuple is a list of our first ten results we specified, the second is some bizarre looking tuple, which if we look at the documentation for ets:select/3 represents something referred to as a continuation.

So we run our query again, and bind the results this time.

{Results, Continuation} = ets:select(TestTable, ProductsInTheTwenties, 10).
% {[{9027,27.99},
%   {7347,29.99},
%   {7282,20.99},
%   {9386,24.99},
%   {5415,25.99},
%   {4032,29.99},
%   {8105,25.99},
%   {4634,24.99},
%   {1275,20.99},
%   {234,20.99}],
%  {16402,576,10,<<>>,[],0}}

So we have this continuation, but what is it and what does it mean for us to have it.

In short, it can be thought of as an immutable bookmark. It represents not only what page we are in for our query results, but also the book we are reading (our query).

This allows us to quickly pick up where we previously left off in our results set by passing the continuation to ets:select/1.

ets:select(Continuation).
% {[{2533,24.99},
%   {1357,22.99},
%   {564,21.99},
%   {9086,22.99},
%   {5265,25.99},
%   {4030,22.99},
%   {2802,25.99},
%   {8254,27.99},
%   {7088,26.99},
%   {3062,27.99}],
%  {16402,960,10,<<>>,[{6792,29.99},{9295,29.99}],2}}

And because it is our special immutable bookmark, every time we use that bookmark it takes us to the same starting point in the same book, and we only read the same number of maximum pages as originally set as our limit.

So no matter how many times we call ets:select/1 with our same continuation, we will get the same results each time.

ets:select(Continuation).
% {[{2533,24.99},
%   {1357,22.99},
%   {564,21.99},
%   {9086,22.99},
%   {5265,25.99},
%   {4030,22.99},
%   {2802,25.99},
%   {8254,27.99},
%   {7088,26.99},
%   {3062,27.99}],
%  {16402,960,10,<<>>,[{6792,29.99},{9295,29.99}],2}}
ets:select(Continuation).
% {[{2533,24.99},
%   {1357,22.99},
%   {564,21.99},
%   {9086,22.99},
%   {5265,25.99},
%   {4030,22.99},
%   {2802,25.99},
%   {8254,27.99},
%   {7088,26.99},
%   {3062,27.99}],
%  {16402,960,10,<<>>,[{6792,29.99},{9295,29.99}],2}}
ets:select(Continuation).
% {[{2533,24.99},
%   {1357,22.99},
%   {564,21.99},
%   {9086,22.99},
%   {5265,25.99},
%   {4030,22.99},
%   {2802,25.99},
%   {8254,27.99},
%   {7088,26.99},
%   {3062,27.99}],
%  {16402,960,10,<<>>,[{6792,29.99},{9295,29.99}],2}}

And if we look at the resulting tuple, we see that we get a different tuple for our next continuation.

{SecondResults, SecondContinuation} = ets:select(Continuation).
% {[{2533,24.99},
%   {1357,22.99},
%   {564,21.99},
%   {9086,22.99},
%   {5265,25.99},
%   {4030,22.99},
%   {2802,25.99},
%   {8254,27.99},
%   {7088,26.99},
%   {3062,27.99}],
%  {16402,960,10,<<>>,[{6792,29.99},{9295,29.99}],2}}

And we can pick up that new continuation, and use that in our next call to ets:select/1 to get the next set of results, with another continuation.

ets:select(SecondContinuation).
% {[{8569,19.99},
%   {1805,28.99},
%   {6819,23.99},
%   {9313,28.99},
%   {9527,27.99},
%   {1737,29.99},
%   {700,26.99},
%   {142,25.99},
%   {6792,29.99},
%   {9295,29.99}],
%  {16402,513,10,<<>>,[],0}}

And if we have a query for which we have exhausted our results set, we get an '$end_of_table' atom.

ets:select(TestTable, [{{'$1', '$2'}, [{'<', '$2', 0}], ['$$']}], 10).
% '$end_of_table'

The ability to specify a limit and have a continuation is also available via on match with ets:match/3 and ets:match/1, and match_object via ets:match_object/3 and ets:match_object/1.

Next week, we will continue looking at the various select functions in ets as we look into their behavior with and ordered set, will look at select vs select_reverse, and play with and see how continuations work if we get some new entries inserted in the results when using a continuation.
–Proctor

One thought on “Erlang Thursday – Using ETS select with a limit

  1. Pingback: Erlang Thursday – ETS selects, continuations, and concurrent insertsProctor It | Proctor It

Leave a Reply

Your email address will not be published. Required fields are marked *