WEBVTT 00:00.120 --> 00:00.690 Hello again! 00:01.050 --> 00:03.360 In this video, we are going to look at tuples. 00:04.020 --> 00:07.350 Tuples in C++ are related to pairs. 00:07.350 --> 00:10.710 So let's first remind ourselves what those do. 00:12.360 --> 00:14.250 A pair is a compound type. 00:14.820 --> 00:19.110 It has two members, which we can access with the names first and second. 00:19.770 --> 00:21.480 These members can have any type. 00:22.320 --> 00:29.100 We can create a pair object explicitly, by giving the types of the elements as template parameters. 00:29.610 --> 00:32.790 So here we have a pair which has two members, of type string. 00:33.830 --> 00:37.410 Or we can use make_pair() to create the pair object, with auto. 00:37.860 --> 00:43.770 So in this case, the compiler will deduce that word_pair is a pair, whose members are both type string. 00:46.190 --> 00:49.550 In C++11, we have a tuple type in the library. 00:50.000 --> 00:51.680 It is defined in the header. 00:52.430 --> 00:55.460 It is similar to the pair, but it can have any number of elements. 00:55.880 --> 00:59.090 The number of elements has to be fixed for a particular object. 01:00.470 --> 01:05.930 It is not really practical to use the names of members, so instead we use the index for accessing the 01:05.930 --> 01:06.380 elements. 01:07.870 --> 01:14.350 Again, we can create a tuple object explicitly, by giving the types of the members as template parameters. 01:15.120 --> 01:18.700 Or we can use make_tuple() and let the compiler deduce it for us. 01:22.360 --> 01:25.750 The way that we access tuple members is a bit peculiar. 01:26.320 --> 01:33.610 There is a non-member function, called get(). This takes the tuple as argumen. And then it has a template 01:33.670 --> 01:36.010 parameter, which is the index of the element. 01:36.880 --> 01:42.850 So if we call get() with parameter zero, that will be the first element. And then we pass the 01:42.850 --> 01:44.500 "numbers" tuple. 01:44.650 --> 01:48.160 So this will return the first element of the numbers tuple. 01:49.300 --> 01:51.790 And we can also assign to the return value from get(). 01:52.990 --> 01:58.660 So we can actually use get() to set the value of an element. Which is perhaps rather confusing naming. 01:59.560 --> 02:02.470 But that is quite common in C++, I'm afraid. 02:03.430 --> 02:09.100 So in this case, we are getting the second element of numbers, and setting its value to three. 02:11.390 --> 02:12.770 In C++14, 02:12.770 --> 02:18.710 we can also use the type of the parameter, provided there is only one elements which has that type. 02:19.550 --> 02:26.330 So we can call get() with parameter int, and that will return the member of numbers which has type int. 02:28.650 --> 02:30.180 So let's try that out. 02:30.660 --> 02:32.280 We create our tuple. 02:32.760 --> 02:39.900 We pass the types of the elements as template parameters. Or we could also calle make_tuple(). 02:41.730 --> 02:44.310 There is our call to get(), with parameter zero. 02:44.820 --> 02:47.730 So this is going to return the first element of this tuple. 02:48.990 --> 02:50.300 get() with parameter 02:50.310 --> 02:56.040 one will be the second member of this tuple. And then we can set that to three. 02:57.510 --> 03:04.320 If we have the parameter type of int, then this will return the member which has type int, which is this 03:04.320 --> 03:04.590 one. 03:05.220 --> 03:07.020 We just set that to three. 03:07.290 --> 03:10.080 So we would expect that this call will return three. 03:11.790 --> 03:12.390 And there we are. 03:13.830 --> 03:14.880 First elements is one. 03:15.270 --> 03:17.040 Second element, set to three. 03:18.630 --> 03:22.440 It is also possible to get all the elements of a tuple with a single call. 03:22.920 --> 03:26.130 There is a non-member function called tie() which does that. 03:26.670 --> 03:28.470 And again, this is a bit peculiar. 03:29.400 --> 03:33.140 We need to have some variables, which we are going to use to store the data. 03:33.750 --> 03:39.600 Then we pass those variables as arguments to tie(), and then we assign the result of calling tie(). 03:39.900 --> 03:45.540 So this is going to extract all the elements from number, and then store the first element in "d", the 03:45.540 --> 03:49.290 second element in "i" and the third elements in "str". 03:49.980 --> 03:55.710 And obviously, the number and type of these variables must match the elements of the tuple. 03:58.010 --> 03:59.130 So let's try that out. 03:59.150 --> 04:00.560 We have the same tuple again. 04:01.340 --> 04:03.800 We have the variables, which we are going to populate. 04:04.460 --> 04:05.390 Then we call tie(). 04:05.420 --> 04:11.360 We pass the variables. And then we assign this to the numbers tuple, and then we print out the result. 04:12.710 --> 04:13.340 So there we are. 04:13.340 --> 04:15.500 The tuple elements are one, two and three. 04:17.650 --> 04:20.290 So when would you want to use a tuple? 04:21.220 --> 04:27.340 One good example is, if you have some data, and you just want to stash it somewhere, without too much trouble. 04:28.090 --> 04:32.920 If you do not wants to write a class and think about member functions, special member functions and 04:32.920 --> 04:36.550 so on. And you are not actually going to use that type anywhere else. 04:37.240 --> 04:40.870 And this is especially useful if the data has elements of different types. 04:41.620 --> 04:45.130 Otherwise you could use a vector, or an array, or some other container. 04:47.380 --> 04:50.830 Another good example is returning multiple values from a function call. 04:51.370 --> 04:53.710 This is a question that I have been asked at interviews. 04:54.490 --> 04:59.050 If you have exactly two values, you could use a pair. But in the general case, you have to create 04:59.050 --> 04:59.530 a struct. 05:00.010 --> 05:02.290 Think up a unique name for it in that namespace. 05:02.770 --> 05:08.440 Then think of names for all the members, then populate all the members in the function body, and then 05:08.440 --> 05:09.070 in the caller. 05:09.340 --> 05:14.920 You have to copy all the data from the members of the struct into your own variables. Which is all 05:14.920 --> 05:15.580 a bit tedious. 05:17.020 --> 05:18.980 With the tuple, you can do it much more conveniently. 05:20.110 --> 05:25.840 You can have a function which returns a tuple, so you just make the parameters the right type for your 05:25.840 --> 05:26.240 data. 05:27.220 --> 05:30.730 You can use a list initializer for creating this tuple object. 05:31.180 --> 05:33.280 So you could do that with variables, as well as literals. 05:34.210 --> 05:40.270 And then, in the caller, you just call tie(). And then this will copy the data, from the returned tuple, 05:40.270 --> 05:41.200 into your variables. 05:42.770 --> 05:49.940 Unfortunately, we cannot use auto, at least not in C++14, because tie() is not able to deduce the types 05:49.940 --> 05:52.070 of the elements. It is just too difficult. 05:54.600 --> 05:55.620 So here is that code. 05:55.650 --> 05:58.110 Here is our function which returns a tuple. 05:58.800 --> 06:01.110 Then we create this tuple using the list 06:01.110 --> 06:01.740 initializer. 06:02.820 --> 06:09.960 Then we have the code which unpacks the tuple into these variables. And we unpack the tuple which was 06:09.960 --> 06:11.010 returned by the function. 06:11.400 --> 06:15.840 So this should give us the data from the function. And there we are. 06:16.440 --> 06:17.900 Okay, so that is it for this video. 06:18.270 --> 06:19.080 I will see you next time. 06:19.080 --> 06:21.360 But until then, keep coding!