Does VS support string_view in C++?

[yungman]

I am playing with string_view which is C++17, I have VS-2019, it doesn't seem to recognize string_view. I cannot even troubleshoot my program:

#include <iostream>
#include <string>
#include <string_view>
using namespace std;
int main()
{
    string_view text = "Hello";
    string_view str(text);
    string_view more(str);
    cout << text << " " << str << " " << more << "\n\n";
    return 0;
}

Thanks.

 

[jedishrfu]

Here’s a write up on it

https://www.modernescpp.com/index.php/c-17-avoid-copying-with-std-string-view

From the examples, it appears that you missing some prefix stuff when using it. The examples use std:: in front of the string_view usage.

 

[yungman]

Here’s a write up on it

https://www.modernescpp.com/index.php/c-17-avoid-copying-with-std-string-view

From the examples, it appears that you missing some prefix stuff when using it. The examples use std:: in front of the string_view usage.

Thanks, I know there must be something wrong with the code. So it's just me, VS is not the issue?

Thanks

 

[Vanadium 50]

it appears that you missing some prefix stuff when using it. The examples use std:: in front of the string_view usage.

His code has "using namespace std", which takes care of it.

However, he doesn't tell us what the error message is, which I think should be the bare minimum for a "debug my code" thread.

 

[Mark44]

See the example in this VS documentation - basic_string_view Class | Microsoft Docs
You need to set a compiler option.

 

[Mark44]

Here's what I did, and I'm running VS 2017.

1. In VS, select the project. For me this is PFTest3.
2. In the right-click menu for the project, at the very bottom, click Properties. This opens the Property Pages dialog for the project.
3. In the left pane, scroll down to C/C++ and then to Command Line under C/C++.
4. In the Additional Options pane near the bottom of this dialog, type in /std:c++17 .
5. Click OK. (I don't show this button in the screen shot below.)

You should then be able to build and run your code.

 

[yungman]

Here's what I did, and I'm running VS 2017.

1. In VS, select the project. For me this is PFTest3.
2. In the right-click menu for the project, at the very bottom, click Properties. This opens the Property Pages dialog for the project.
3. In the left pane, scroll down to C/C++ and then to Command Line under C/C++.
4. In the Additional Options pane near the bottom of this dialog, type in /std:c++17 .
5. Click OK. (I don't show this button in the screen shot below.)

You should then be able to build and run your code.
View attachment 276866

Thanks so much Mark44. It works My program compiles.

I am really not into studying string_view. The Ivor book gave an example using string_view and I went nowhere on this. I only made up the little program to confirm it's VS.

What is that you are doing?

I was playing with this, you can actually click Project on the top bar, choose the last line PFTest3 properties and do the same thing.

Thanks so much,

 

[yungman]

This is the program that I am trying to understand and want to translate to using c-string so I can understand it better:

#include <iostream>
#include <cstring>
#include <string>
#include <string_view>
using namespace std;
class Trouble
{
private: string message;
public:
    Trouble(string_view str = " There is a problem") { message=str; }
    string_view what() const { return message; }
};
void trySomething(int i)
{
    if (i == 0) throw Trouble();
    else throw Trouble{ " Nobody knows the trouble I've seen..." };
}

int main()
{
    for (int i = 0; i < 2; i++)
    {
        try { trySomething(i); }
        catch (const Trouble& t) { cout << " Exception: " << t.what() << "\n\n"; }
    }
    return 0;
}

Still working on it, it's from Ivor book.

 

[yungman]

Please help me on this. I am translating the program in post#8 to use only c-string instead.

After I translate the program, it only print out the first letter of the string literal. eg, the first sentence is only "T" instead of "There is a problem". See output of program below.

I verified the way I pass the string literal to a function shown in line 20 to 23 and got the whole sentence "This is OutCstring test".

#include <iostream>
#include <cstring>
using namespace std;
const int csize = 51;
class Trouble
{ public:
    char message[csize];
    Trouble(const char *str)//Declare str pointer
        {
        strncpy_s(message, csize, str, csize);
        cout << " message: " << message << "\n\n";//Print out "There is a problem" first
        }
    char what() const  { return *message; }
};
void trySomething(int i)
{
    if (i == 0) throw Trouble("There is a problem");//Only pass the first letter "T"
    else throw Trouble( "Nobody knows the trouble I've seen..." );//Only pass letter "N"
}
void OutCstring(const char* C)//To show it passes the string literal
{
    cout << " The c-string passed to OutCstring is: " << C << "\n\n";
}

int main()
{
    OutCstring(" This is OutCstring test");//To proof I pass the whole sentence.

    for (int i = 0; i < 2; i++)
    {
        try { trySomething(i); }
        catch (const Trouble& t) { cout << " Exception: " << t.what() << "\n\n"; }//Only pass first letter.
    }
    return 0;
}

This is the output:

The c-string passed to OutCstring is: This is OutCstring test// test passing string literal to function

message: There is a problem //Print in line 11 to show message contains “There is a problem”

Exception: T //Only the first letter of the sentence “There is a problem”

message: Nobody knows the trouble I've seen... //Print in line 11 show it got copied correctly

Exception: N //Only the first letter of “Nobody knows the trouble I’ve seen……

I don't know what went wrong, please help
Thanks

 

[Filip Larsen]

I don't know what went wrong, please help

Try look at the return type for your what() method.

 

[yungman]

Try look at the return type for your what() method.

Thanks

I kind of thinking it's the return type, I should return the message, but if I remove the '*', it flag me an error. returning *message will give the content of the address of the first element of the c-string message which is the 'T'. But I don't know how to fix it. I try everything with that line already.

Thanks

 

[Filip Larsen]

I kind of thinking it's the return type

It is so basic I wonder how you cannot spot it yourself. You have declared the what() method to return a char. Does this match any kind of string type? If no bell rings yet then try to look at one of the many examples involving C-strings you already have discussed here on PF.

 

[Vanadium 50]

but if I remove the '*', it flag me an error

@yungman, you really, really, really need to go back and review the section on pointers. I know you've told us you know it. But you really need to learn it better. Randomly adding/removing *'s and &'s hoping it will hope shows that you don't have this down.

And getting back to the topic at hand, the reason it didn't work for you is because you need to tell VS you are using c++17. That's what @Mark44 does in his message.

 

[Filip Larsen]

the reason it didn't work for you is because you need to tell VS you are using c++17.

Since yungman indicate the code compiles and runs (but with, for him, unexpected results), I will maintain my earlier claim that a necessary condition to the what() to return a string is to change the return type to a string.

 

[Vanadium 50]

Since yungman indicate the code compiles and runs

Did he? All he said was "I cannot even troubleshoot my program"

 

[Filip Larsen]

Did he? All he said was "I cannot even troubleshoot my program"

Ah, missed that you replied to the first post.
My reply is relevant for the code and question in post #8, which is the latest code sample (both now and when I replied).

 

[yungman]

@yungman, you really, really, really need to go back and review the section on pointers. I know you've told us you know it. But you really need to learn it better. Randomly adding/removing *'s and &'s hoping it will hope shows that you don't have this down.

And getting back to the topic at hand, the reason it didn't work for you is because you need to tell VS you are using c++17. That's what @Mark44 does in his message.

That is out of desperation because it doesn't make sense. I review my notes over, I NEVER managed to return a c-string from a function. I did asked here and I never got an answer, so I had to conclude I CANNOT return a c-string like that. If you know of a way, let me know.

Yes, I got that it's the VS from the very start already, that's why I asked in the first place. The error showed that it's obvious there's nothing wrong with the program, VS just won't compile without doing something. Never know VS needed to be told that it's C++17.

 

[yungman]

I came up with a different way to return the message.

#include <iostream>
#include <cstring>
using namespace std;
const int csize = 51;
class Trouble
{ public:
    char message[csize];
    Trouble(const char *str)//Declare str pointer
        {     strncpy_s(message, csize, str, csize);     }
    ~Trouble() { cout << " In Trouble destructor.\n\n"; }
//    char what() const  { return *message; }
};
void trySomething(int i)
{
    if (i == 0) throw Trouble("There is a problem");
    else throw Trouble( "Nobody knows the trouble I've seen..." );
}
int main()
{
    for (int i = 0; i < 2; i++)
    {
        try { trySomething(i); }
        catch (const Trouble& t) { cout << " Exception: " << t.message << "\n\n"; }//just get the message directly!
    }
    return 0;
}

I just simply use t.message to get the message out. Easy peesy! I eliminated all the extra, but I added in the destructor to look at when it's out of scope.BTW, I did asked about function returning c-string here. I could not find anything in the book, I did not get any reply here. I search the web also. So I never have that in my notes. I concluded that it's NOT possible to return c-string from function. If you guys have a way, please let me know.

Also, seems like the Ivor books is the ONLY one that go deep into this exception. The C++ Primer book by Lippman arrived, it only has like 5 pages on exception, even less than Gaddis. Is exception that important? Why only Ivor put in so much effort in this? I finished Gaddis 20 pages in less than a week. All the time and debate are from the Ivor book. If exception is not that important, I am going to stop here. I already learn a lot more than in Gaddis book already. I want to move onto Template.

Thanks

Ah, missed that you replied to the first post.
My reply is relevant for the code and question in post #8, which is the latest code sample (both now and when I replied).

That's two completely different questions. My first question in OP is about VS, that was resolved thanks to Mark44. I just tagged on the second question in post 8. Vanadium50 mixed the two together.

 

[Filip Larsen]

I did asked about function returning c-string here. I could not find anything in the book, I did not get any reply here. I search the web also.

Really?

First hit when searching for "c++ function returning c-string" (using duckduckgo):
https://stackoverflow.com/questions/1496313/returning-a-c-string-from-a-function

You could also try "what is a c-string" and see if that brings more light to your understanding of this.

The literature and the web is simply stock full of char * and const char *, and I have a hard time imagining how you could getting this far in your C++ exercises without never noticing what a C-string is. As you have been told by many by now, it is very difficult to assist you with appropriate hints when you "suddenly" are completely blank on some of the most basic concept.

 

[Vanadium 50]

I NEVER managed to return a c-string from a function.

That is correct, and that is why I suggest you go over pointers again. A function cannot return a string. It can return a pointer to a string.

 

[yungman]

Really?

First hit when searching for "c++ function returning c-string" (using duckduckgo):
https://stackoverflow.com/questions/1496313/returning-a-c-string-from-a-function

You could also try "what is a c-string" and see if that brings more light to your understanding of this.

The literature and the web is simply stock full of char * and const char *, and I have a hard time imagining how you could getting this far in your C++ exercises without never noticing what a C-string is. As you have been told by many by now, it is very difficult to assist you with appropriate hints when you "suddenly" are completely blank on some of the most basic concept.

I tried before, I can't find a way to make it work. This is the program I was working on, you can see the one in the middle doesn't work.

#include <iostream>
#include <cstring>
using namespace std;
const int csize = 21;
const char* retCstring1() 
    {
        return "Alan";
    }
const char* retCstring2()
    {    
        const char Cret[] = "Paul";
        return Cret;
}
void passCstring(char* C)
    {    
        strncpy_s(C, csize, "John", csize);
    }

int main()
{    int a;    char Cr[csize];
    strncpy_s(Cr, csize, retCstring1(), csize);
    cout << " Return from retCstring1: " << Cr << "\n\n";
    //strncpy_s(Cr, csize, retCstring2(), csize);
    cout << " Return from retCstring2: " << retCstring2() << "\n\n";//wont work.
    passCstring(Cr);
    cout << " passing from passCstring: Cr = " << Cr << "\n\n";    
    return 0;
}

The output is:

 

[Vanadium 50]

You should also review functions. A function that returns a string would need to have a string type. But C does not have a string type. (It does. however, have a pointer type)

 

[yungman]

You should also review functions. A function that returns a string would need to have a string type. But C does not have a string type. (It does. however, have a pointer type)

That's what I am talking about. I failed to see a way with this. I spent a lot of time on this. I know I can work around like in my final program bypass that. That's easy. My question is specifically on return c-string from a function.

I'll be really thankful if you can help making the middle one work. That's the essence of return a c-string. I really worked on it. I got so desperate I put '*' and const blindly all over already.

Hey, you can be logical, BUT if none work, the NEXT logical step is to go illogical!:)

 

[Vanadium 50]

My question is specifically on return c-string from a function.

And you cannot do that. I've told you twice, Filip has told you once, and that's this thread. In your previous thread, Mark44 tells you twice, Jarvis323 tells you twice, and Halc and sysprog each tell you once. And if I searched more threads, I would find more examples.

It's disrespectful to us and wasteful of our time when we answer a question that you've asked, and you don't read the responses, and ask the same question again.

You really need to go back to the chapters on functions and pointers.

 

[Vanadium 50]

To rehash and summarize:

1. A function cannot return a C-string in C. (Or C++)
2. A function can return a pointer to a C-string. However, one must take care to ensure this pointer points to the string, and not a memory location where the string used to be.

 

[jbunniii]

I tried before, I can't find a way to make it work. This is the program I was working on, you can see the one in the middle doesn't work.

const char* retCstring2()
    {
        const char Cret[] = "Paul";
        return Cret;
}

That's because retCstring2() is returning a dangling pointer. The array Cret[] is a local variable which goes out of scope as soon as the function returns. So, the pointer returned by retCstring2() points to memory that is no longer available.

The result is undefined behavior. If you're lucky the memory won't yet be overwritten when you try to use the pointer returned by retCstring2() , but in general this is not guaranteed. In the real world, Murphy's Law means that it will work when you test it, you'll ship it to customers, and it won't work for them.

As I've mentioned in a previous message in another thread, the only way to do what you're trying to do is to allocate an array of storage using new, copy the string into that array, and return a pointer to that array.

But then the burden falls on the caller of your function to delete the storage. This leads to all sorts of memory leaks when users fail to do so. The natural thing to do is to create a class that performs the new in the constructor and the delete in the destructor, hence cleans up after itself. But this is just a very poor duplication of what std::string already does, so why not use std::string?

The only other alternative is something like your Trouble implementation, which uses a fixed-size array and thereby avoids the new/delete issue. But this is a poor solution because it imposes a maximum string length (the size of the array must be a constant known at compile time), and every instance of this class will require the maximum fixed storage size, even if only using a tiny part of it!

class Trouble
{ public:
    char message[csize];
    Trouble(const char *str)//Declare str pointer
        {     strncpy_s(message, csize, str, csize);     }
    ~Trouble() { cout << " In Trouble destructor.\n\n"; }
//    char what() const  { return *message; }
};

No one should be messing around with C strings in 2021 unless they have a very good reason, such as (1) they're on a compiler team implementing the standard library, hence have to write std::string from scratch, or (2) they're on a primitive embedded platform that doesn't provide dynamically allocated memory and therefore cannot provide a std::string implementation.

 

[pbuk]

This is absolutely golden advice:

No one should be messing around with C strings in 2021 unless they have a very good reason, such as (1) they're on a compiler team implementing the standard library, hence have to write std::string from scratch, or (2) they're on a primitive embedded platform that doesn't provide dynamically allocated memory and therefore cannot provide a std::string implementation.

In the real world, we write exceptions in C++ to throw objects with integer (enum) error codes. If a human language message needs to be displayed/logged this can be dealt with in the error handling block of the calling code.

 

[Filip Larsen]

No one should be messing around with C strings in 2021

I have been saying that from the start too! But someone here keep insisting on continuously experiencing and share how easy is it to shoot yourself in the foot (or other appropriate body part) with these strings .

(For the record, I give said someone plus points for stamina, but minus point for not learning from earlier mistakes - and myself minus points for keep falling into the trap of answering ).

 

[yungman]

I have been saying that from the start too! But someone here keep insisting on continuously experiencing and share how easy is it to shoot yourself in the foot (or other appropriate body part) with these strings .

(For the record, I give said someone plus points for stamina, but minus point for not learning from earlier mistakes - and myself minus points for keep falling into the trap of answering ).

The reason is Gaddis book mainly use c-string for everything. I am still using the Gaddis and do the exercise. Gaddis book still have the best program examples, even have CD that contain all the programs. I cannot just drop it. I have no choice at this point! After learning exception,Templates and the few chapters on data structure, then if I have time, I can go back in the future, but not now.

Yes, I notice Ivor book and the newly acquired C++ Primer by Lippman both exclusively using std::string. Lippman book is better than Ivor, BUT it really doesn't have complete program that I can work on, just snipped of codes. Ivor is hard to understand period, we went through a big debate on that already. Lippman book is better for reading. So I am still confined to reading Lippman and working on Gaddis...Meaning a lot of c-string.

Also, c-string is faster in run time than std::string. I have to learn string_view of C++17 to speed it up. That's one more to learn for me where I am trying to finish Gaddis book.

Yes, I notice std::string is much easier to use.

 

[Filip Larsen]

I notice Ivor book and the newly acquired C++ Primer by Lippman both exclusively using std::string. Lippman book is better than Ivor, BUT it really doesn't have complete program that I can work on, just snipped of code.

I am not familiar with Lippman's book, but hopefully you should quickly be able to work those snippets into your own context pretty quickly (just like you seem to do with the code you have from Gaddis).

c-string is faster in run time than std::string.

Even if that claim was true (which I strongly doubt; modern std::string or similar implementations are highly optimized for performance while still maintaining easy and safe use), I can't help wondering what kind of program you plan to make where you think any performance difference between the two will be an issue?

I ask because run-time performance obviously does not matter one bit when you are just learning the basics, so if you are worried about speed you must be thinking of some usage situation and not about learning, right?

 

[yungman]

Even if that claim was true (which I strongly doubt; modern std::string or similar implementations are highly optimized for performance while still maintaining easy and safe use), I can't help wondering what kind of program you plan to make where you think any performance difference between the two will be an issue?

I ask because run-time performance obviously does not matter one bit when you are just learning the basics, so if you are worried about speed you must be thinking of some usage situation and not about learning, right?

I read it on line to justify going into string_view.

I hope I don't offend you below:

I designed a lot of hardware, there are situations time is critical. That goes back to a lot of times I mentioned and people disagree here. I notice EVERY single piece of hardware/appliance, be it cars, tv, computers, printers...everything that using firmware and/or software are getting slower and slower while the speed of hardware and processors are getting faster. I blame it on software people taking for granted that hardware is fast, they they can be more fancy and don't worry about the speed. I just do NOT want to be one of them. I respect hardware people that bent over backwards, breaking their backs to squeeze out the last bit of speed just to be wasted and take for granted by others. Do you know how hard it is to squeeze the speed, generation of hardware engineer, pcb designers are forced out because they don't have the knowledge of microwave to do design? The bus system on the pcb is NOT just a trace, you have to worry about reflection, crosstalk and all that.

I know how to design these, I spent years studying calculus, ODE, PDE in order to study electromagnetic to design microwave hardware. I few sorry towards those that don't. Out of respect, I will NOT wasting the speed away no matter what. There is a lot of blood sweat and tears to this.

 

[Filip Larsen]

I designed a lot of hardware, there are situations time is critical.

Granted, on (small or special) embedded devices you may have limited resource and the programming environment may also be fairly limited (e.g. no heap and similar), but then I would think you would probably also be coding more in C than "full" C++ (and C-strings obviously makes sense to learn about if you want to code in C).

I would still venture the guess that you would be hard pressed to find an embedded device with accompanying SDK supporting full C++ but only in such a way that resource managed strings (like std::string or its replacement on that particular platform) are significantly slower than "unmanaged" raw string pointers. My point here is that if you have C++ you also have the means to encapsulate the complexities for handling raw string pointer in a class (whether they are null terminated or length based) without being forced to loose performance.

Also, if "string" processing performance is critical on an embedded CPU device, perhaps some of it should be handled by a DSP or a FPGA instead. For instance, at my work we do mostly C++ with a dash of OpenCL code that is deployed to data processing GPU's and special FPGA's. Managing buffers in that context is similar to C-string handling, but even so we have handy C++ classes to wrap all the lower-level stuff.

 

[jbunniii]

Even if that claim was true (which I strongly doubt; modern std::string or similar implementations are highly optimized for performance while still maintaining easy and safe use), I can't help wondering what kind of program you plan to make where you think any performance difference between the two will be an issue?

Agree strongly with this. Any modern compiler will have a highly optimized implementation of std::string and other standard library features. (Additionally, they have even-more-optimized specializations for small strings which require almost no overhead beyond the storage for the characters themselves.) It's a premature pessimization to assume otherwise, especially if it means trading the safety and convenience of std::string for the terribly error-prone and limited C string. The latter is hardly suitable for high performance without extra bookkeeping anyway. Even simple operations like determining the length of a C string (strlen) require iterating through the entire string to look for the terminating '\0'.

Moreover, @yungman, I strongly suspect that in most/all of the use cases for which you are considering std::string_view, you could achieve the equivalent by simply passing a std::string by reference instead of by copy. Can you show an example where you think this is not the case?

 

[Filip Larsen]

I notice std::string is much easier to use.

Indeed, and that ease of use goes directly into better code quality, more flexible code and higher coding speed (and more fun doing all that)

 

[Jarvis323]

I read it on line to justify going into string_view.

I hope I don't offend you below:

I designed a lot of hardware, there are situations time is critical. That goes back to a lot of times I mentioned and people disagree here. I notice EVERY single piece of hardware/appliance, be it cars, tv, computers, printers...everything that using firmware and/or software are getting slower and slower while the speed of hardware and processors are getting faster. I blame it on software people taking for granted that hardware is fast, they they can be more fancy and don't worry about the speed. I just do NOT want to be one of them. I respect hardware people that bent over backwards, breaking their backs to squeeze out the last bit of speed just to be wasted and take for granted by others. Do you know how hard it is to squeeze the speed, generation of hardware engineer, pcb designers are forced out because they don't have the knowledge of microwave to do design? The bus system on the pcb is NOT just a trace, you have to worry about reflection, crosstalk and all that.

I know how to design these, I spent years studying calculus, ODE, PDE in order to study electromagnetic to design microwave hardware. I few sorry towards those that don't. Out of respect, I will NOT wasting the speed away no matter what. There is a lot of blood sweat and tears to this.

The most important thing is to know what you're doing first. If you did have a use case where c-strings performed better than std::string, would it matter anyway if your code is wrong, crashes randomly, hangs, gives wrong results etc.

I suggest that you should first figure out what a c-string is and what an std::string is, how to use them, and then figure out in which cases one might be faster than the other, and by how much, by testing the performance.

It's always important to know what you are doing, both when it comes to programming logic, syntax, etc., and optimization.