WEBVTT 00:02.580 --> 00:08.160 Hello again! In this video, we're going to look at if statements and switch in C++ 17. 00:10.540 --> 00:17.200 Let's start off with something that we already know about. In C and early versions of C++ if you want 00:17.200 --> 00:18.070 to write a for loop, 00:18.370 --> 00:21.400 you have to declare the loop counter before the loop. 00:22.120 --> 00:29.530 So this means that this variable i, which we are going to use to iterate over - some of array, presumably - is 00:29.530 --> 00:31.510 actually defined in the enclosing scope. 00:32.950 --> 00:36.970 So after the loop has finished, this variable i will still be in existence. 00:37.480 --> 00:39.880 We can't create another variable called i. 00:40.780 --> 00:43.780 And if you want another loop, you either have to use a different variable 00:44.320 --> 00:49.720 Or we have to re-use this variable i, which in itself can create bugs. 00:49.990 --> 00:54.700 If someone else has decided to use that variable and then we change the value to something that they were not 00:54.700 --> 00:55.270 expecting. 00:56.780 --> 01:04.620 In C++ 98, you now allowed to declare the variable inside the for loop, so this is a for loop initializer. 01:04.640 --> 01:12.830 i is now a local variable inside the loop and once the loop is complete, then the variable i no longer 01:12.830 --> 01:13.370 exists. 01:14.780 --> 01:18.500 And if you want to write another loop, you can just do the same thing, "for (int i ..." 01:19.400 --> 01:21.220 And you will get your own version of i. 01:24.160 --> 01:30.670 So this is really a better way of doing things, because the loop counter is really an internal detail 01:30.670 --> 01:36.340 of the loop. Code outside the deep should not need to know what type the loop counter is, what its name 01:36.340 --> 01:37.330 is, what its value is. 01:38.230 --> 01:43.720 If you do need to know the final value of the loop counter, you can always define a variable outside the 01:43.720 --> 01:45.070 loop and copy into it. 01:47.830 --> 01:53.170 So this makes your code clearer, more concise and safer. Which is all good stuff! 01:59.750 --> 02:05.300 We often have situations where we want to check the return value from a function before we do something. 02:05.870 --> 02:12.170 So this normally means we have to define a variable to hold the returned value and then write an statement. 02:12.920 --> 02:14.660 But really, this is just a single operation. 02:14.660 --> 02:17.480 If the return value is so and so, do this. 02:20.360 --> 02:26.030 In this example, we are calling begin() to get the first element of a vector, and we are using this variable 02:26.030 --> 02:31.220 "iter" to store the result, and then we want to make sure that the vector is not empty. 02:31.580 --> 02:35.900 If the vector is empty, then the results of calling begin() is the same as the results of calling end(). 02:36.950 --> 02:42.860 So we want to check the vector is not empty before we process it. In C++17, 02:42.860 --> 02:44.570 you can do this all in a single statement. 02:44.570 --> 02:47.120 You can initialize the iterator. 02:47.630 --> 02:54.200 So we have this initialization statement here and then semicolon and then we have our condition. 02:55.520 --> 03:01.310 So this defines iter as a local variable to the if statement, and then it does the check. 03:01.610 --> 03:05.180 And if this test is true, then it will execute the body of the if statement. 03:05.540 --> 03:10.670 And this has similar advantages to the loop initializer. It does not get mixed up with the rest of 03:10.670 --> 03:11.060 the code. 03:12.440 --> 03:18.860 When we say that iter is local to the if statement, that also includes an "else" block, which corresponds to 03:18.860 --> 03:19.290 the if. 03:20.180 --> 03:25.760 So in this case, it's not really very useful because the else statement will only happen if the iterator 03:25.760 --> 03:28.520 is invalid, which means you cannot do anything useful with it. 03:29.210 --> 03:32.600 But in other situations, with a normal variable, that might be useful. 03:34.600 --> 03:39.910 So here is that code in Visual Studio. We have our vector. 03:40.960 --> 03:47.500 Then the old way. We get the iterator to the first element, then we compare it to the result of 03:47.500 --> 03:48.100 calling end(). 03:49.000 --> 03:52.890 And if that is different, then it is safe to use the iterator. 03:54.490 --> 04:00.700 Then we have the C++17 version, where we initialize the iterator inside the if statement. 04:01.480 --> 04:02.800 So this will call begin(), 04:03.220 --> 04:08.320 store the result and then compare it to the result of calling end. Unfortunately, 04:10.290 --> 04:17.760 Visual C++ doesn't actually support this. "Requires C++ 17", but if you look in the project properties, 04:20.090 --> 04:21.990 we are using C++ 17! 04:22.010 --> 04:27.590 So there you are, four years after C++ 17, and it cannot - and it is still not fully supported. 04:28.070 --> 04:31.430 People often ask me why I do not cover C++ 20 in my courses. 04:31.940 --> 04:38.900 There is the reason. We still do not have complete support for C++ 17. And C++ 20 is obviously much less 04:38.900 --> 04:39.410 complete. 04:41.450 --> 04:50.600 So I am going to use MinGW instead, I use Notepad++ and MinGW for writing code for videos, on Windows. 04:51.080 --> 04:54.350 I only use Visual Studio for demonstrations, because it is a bit easier. 04:55.340 --> 04:56.750 So this is MinGW. 04:56.750 --> 05:00.230 This is a port of GNU C++ for Windows. 05:00.860 --> 05:07.320 The editor is Notepad++, which is a free program. Not the Notepad that comes with windows! 05:07.340 --> 05:10.370 This is quite different. And infinitely better! 05:12.460 --> 05:13.930 So I have got the same code here. 05:14.260 --> 05:15.690 I am not sure if you can actually see that, 05:15.820 --> 05:18.160 because it is a bit smaller, but it is actually the same. Honest! 05:26.100 --> 05:30.420 So that's how you get C++ 17 with GCC and also clang. 05:31.380 --> 05:32.520 And then the name of the file. 05:35.350 --> 05:41.440 And right, so it does actually know how to compile C++ 17 code. And then we can execute it. 05:41.530 --> 05:46.300 The output file is called "a.exe", unless you ask for it to be called something else. 05:49.970 --> 05:53.570 So there we are, we get the same resultswith both forms of the code. 05:59.020 --> 06:04.900 And you can also do this with a switch statement. So instead of creating a variable to store the result 06:04.900 --> 06:09.460 of the function call and then switching on that, you can do it all in a single statement. 06:12.130 --> 06:15.310 So here's a simple program which uses a switch statements. 06:16.180 --> 06:20.260 I mean, this is really C code, iterating over array with an integer. 06:20.380 --> 06:27.400 So perhaps it's not the best example of Modern C++! But we do have this C++ 17 feature with the initializer 06:27.400 --> 06:28.180 in the switch statement. 06:29.770 --> 06:35.470 So we have this C-style string with some whitespace characters. We have some spaces and also 06:35.470 --> 06:37.420 this tab character. 06:38.620 --> 06:44.080 And the point of this program is to count the number of spaces in this string. 06:44.890 --> 06:49.270 We have this variable here that we're going to use to count the number of whitespaces. 06:49.510 --> 06:53.140 So every time we encounter a whitespace character, we're going to increment it. 06:55.890 --> 06:58.080 And then we iterate over this array. 06:58.110 --> 07:03.630 And then for each element, we gets the current character and we have a switch. 07:05.190 --> 07:09.870 So let's remind ourselves how switch statements work. When the program runs, the 07:11.100 --> 07:19.050 program will examine this character 'c', so this is some numeric value, and then it will jump to the case label that matches it. 07:19.470 --> 07:25.860 So if C is a space character, it will jump to this case, if it is a tab, it will jump to here and if it is a newline, it will 07:25.860 --> 07:26.460 jump to here. 07:27.570 --> 07:32.580 And then once it has jumped to a case, it will execute all the code until it encounters a break statement. 07:33.330 --> 07:36.720 So it can actually execute across case labels. 07:38.490 --> 07:44.100 So this means if c is a space or a tab or a new line, it is going to execute the same code, which is to 07:44.100 --> 07:45.060 increment the counter. 07:45.600 --> 07:50.400 And then when it gets to the break, it will jump out of the switch statement and go on to the next iteration 07:50.400 --> 07:51.000 through the loop. 07:52.080 --> 07:59.430 If it is anything other than a space, tab or newline, it is going to jump to the default label. Which will 07:59.430 --> 08:02.190 just jump out of the switch again and go to the next element. 08:05.340 --> 08:09.300 So again, Visual C++ will not compile this code, so we have to get back to... 08:19.870 --> 08:22.930 So there we are. The text is, "How much white space in here?" 08:23.560 --> 08:27.130 And the answer is five, so one, two, three. 08:27.810 --> 08:28.350 Four. 08:29.990 --> 08:30.380 Five. 08:30.680 --> 08:33.680 Yes, there is an extra space after the tab. That confused me! 08:38.240 --> 08:43.160 So we say in this case that the code will fall through the case labels until it finds a break statement. 08:44.690 --> 08:48.050 There were other languages which work differently. In some languages, 08:48.050 --> 08:50.150 only one case statement is ever executed. 08:50.480 --> 08:55.870 So if you have an empty case that would presumably be an error. But in C++, it just keeps falling through 08:55.880 --> 08:57.830 until it finds a break statement. 08:59.300 --> 09:01.460 And this can be quite useful as in this program. 09:02.240 --> 09:07.700 On the other hand, if you make a mistake and forget to put in a break statement, then it can be rather 09:08.090 --> 09:08.420 bad. 09:10.450 --> 09:17.920 So in this example, we have people who work in different locations in the company, and we want to 09:17.920 --> 09:19.930 assign them a car, depending on where they work. 09:20.500 --> 09:26.830 So we have this employee location, which is presumably some sort of enumerated type, and we do a switch 09:26.830 --> 09:27.130 on it. 09:27.790 --> 09:34.780 So if they work in head office, they get a nice swanky car. If they work in sales they get a fairly average 09:34.780 --> 09:35.110 car. 09:35.980 --> 09:38.650 And if they work in the warehouse, they get to drive a truck. 09:40.300 --> 09:41.710 So that's the theory anyway. 09:41.710 --> 09:44.320 But the programmer forgot to put a break statement in. 09:44.650 --> 09:50.200 So when this runs, if the employee location is head office, then it's going to set the vehicle type 09:50.200 --> 09:56.020 to Mercedes. Then it's going to fall through to sales. Then it wil set the vehicle type to Peugeot. Then it will fall 09:56.020 --> 09:58.030 through and set the vehicle type to truck. 09:58.690 --> 10:04.060 So in fact, every employee will end up driving a truck, which is probably not going to make you very 10:04.060 --> 10:04.480 popular! 10:08.170 --> 10:12.850 Some compilers will actually give you a warning here, which is quite helpful because it tells you that 10:12.850 --> 10:18.400 you may need to check program logic. On the other hand, if you did actually intend to fall through, 10:18.730 --> 10:19.420 then it is annoying. 10:19.720 --> 10:21.730 It is just more noise in your compiler output. 10:22.420 --> 10:27.130 And if you have a setting which treats all warnings as errors. Or if you have a coding standard 10:27.130 --> 10:31.840 which says you are not allowed to have any warnings at all, then it is very annoying! 10:32.680 --> 10:36.580 What usually happens in that case is people just get annoyed and hack their code to shut the compiler up, 10:36.880 --> 10:38.530 which is not really what we want to happen. 10:42.300 --> 10:44.730 C++ has things called attributes. 10:44.820 --> 10:47.100 These are nothing like Java attributes! 10:47.460 --> 10:52.920 You can't go and implement a database query system with them. An attribute is just simply an instruction 10:52.920 --> 10:55.770 to the compiler, and it cannot change the meaning of the code. 10:59.150 --> 11:04.760 To write an attribute, we put it inside a pair of square brackets, and the language defines quite 11:04.760 --> 11:05.360 a few of these. 11:05.660 --> 11:06.740 I'll be going through all of them. 11:07.100 --> 11:11.960 But as we are talking about switches and C++ 17, this seems like a good point to cover it. 11:13.880 --> 11:19.940 So to use the fallthrough attribute, we put it in an empty statement. So that just means we put a semicolon 11:19.940 --> 11:25.390 after it, like that. And that will tell the compiler that we mean to fall through and it is not a mistake 11:25.400 --> 11:26.930 and please do not give a warning. 11:31.870 --> 11:36.820 So here is the code in visual C++ again. Which again, does not compile. 11:48.860 --> 11:49.460 So we are. 11:50.750 --> 11:52.640 The code compiles without any warnings. 11:52.880 --> 11:55.910 So if we - Let's say we try to make it warn on everything. 12:02.280 --> 12:07.410 OK, we get a warning for using an unused variable, but we do not get the [warning] for falling 12:07.410 --> 12:09.450 through, which is what we are trying to avoid. 12:12.550 --> 12:14.170 OK, so that's it for this video. 12:14.680 --> 12:17.560 I'll see you next time, but meanwhile, keep coding!