C / C++ / MFC

Re: Class as memebrs of other classes : Create a constructor that takes all data

6-Apr-19 9:40

To answer your question

C++

class book {
  ...
  book(string title, int year, string name, string email) : auth(name, email) {
  ...
  }
};

Some things for you to think about:
1) If you have getters and setters for all your private members, how is this different than making the private members public?
2) Although names don't usually change change (Ferdinand Lewis Alcindor Jr might disagree), emails can and do. What happens if you have a large collection of books and want to change the email of an author? There might be a better way to structure your data.

Stefan_Lang8-Apr-19 21:51

Stefan_Lang

8-Apr-19 21:51

@1) he doesn't have a getter and setter for book::author Wink | ;-)

@2) agreed. Either the class author should really be a class author_reference, and name and email are just a (n unsuitable) choice of a unique reference to this specific author which is represented within a separate class and my contain more (private) data like home address and preferred brand of peanut butter. Or the class author really represents an actual author, in which case (s)he should only be referenced, not defined as a data member. After all, the author shouldn't die just because a book is burned... OMG | :OMG:

GOTOs are a bit like wire coat hangers: they tend to breed in the darkness, such that where there once were few, eventually there are many, and the program's architecture collapses beneath them. (Fran Poretto)

Re: Class as memebrs of other classes : Create a constructor that takes all data

Gerry Schmitz6-Apr-19 10:36

Gerry Schmitz

6-Apr-19 10:36

Besides an author having multiple books, a book can also have multiple authors (including commentators).

"(I) am amazed to see myself here rather than there ... now rather than then".
― Blaise Pascal

Re: Class as memebrs of other classes : Create a constructor that takes all data

jschell7-Apr-19 8:53

jschell

7-Apr-19 8:53

Presumably this is homework.

At any rate a "copy constructor" sounds quite possibly what the homework really wants you to come up with. Although perhaps the assignment is to let you copy everything first and then next demonstrate what a copy constructor is.

New [[nodiscard]] attribute usage.

Maximilien4-Apr-19 5:57

Maximilien

4-Apr-19 5:57

The [[nodiscard]] attribute can be use either for function return and for struct.

Both will return the same warning; Is there a better use case for the attribute ?

Either tag the struct:

C++

struct [[nodiscard]] Patate
{
    int patate;
};

Patate f1()
{
    return Patate();
};

void main()
{
    f1(); // warning C4834: discarding return value of function with 'nodiscard' attribute.
}

or tag the function return.

C++

struct Patate
{
    int patate;
};

[[nodiscard]] Patate f1()
{
    return Patate(); 
};

void main()
{
    f1(); // warning C4834: discarding return value of function with 'nodiscard' attribute.
}

I'd rather be phishing!

John R. Shaw4-Apr-19 7:56

John R. Shaw

4-Apr-19 7:56

Your function is returning by value, which your code is ignoring (discarding).
A Google search for "attribute nodiscard" provides the answer.

C++ attribute: nodiscard (since C++17) - cppreference.com[^]

INTP
"Program testing can be used to show the presence of bugs, but never to show their absence." - Edsger Dijkstra
"I have never been lost, but I will admit to being confused for several weeks. " - Daniel Boone

Maximilien4-Apr-19 16:17

Maximilien

4-Apr-19 16:17

I know that, but the description is not clear about the differences if any between the nodiscard for the struct vs for the function.

I'd rather be phishing!

Randor 4-Apr-19 18:53

Randor

4-Apr-19 18:53

Well,

That attribute is weakly defined in section 10.6.7 of the standard. You can go read it yourself. The behavior is intentionally vague using words like 'encouraged' and 'discouraged'. Regardless of whether it's applied to a function or class there is no difference defined.

In fact if you carefully read the document no behavior has been defined. The only section without 'encouraged' and 'discouraged' is section 1 where the attribute token "[[nodiscard]]" itself is defined.

In other words in C++17 the only guarantee is that your compiler understand the [[nodiscard]] attribute. No behavior is defined.

Best Wishes,
-David Delaune

leon de boer5-Apr-19 3:24

leon de boer

5-Apr-19 3:24

There is no difference you are discarding something (you didn't save it) that was marked as non discard.

The whole idea should have been euthanized, it is what you get when you get a committee to talk about non existent problems.
There are half a dozen real problems but they waste time on that junk.

Here is the best template to fix [nodiscard]

template < typename T >
void ignore(T&&) {}

Try it on your code

ignore(f1());

Now don't make me do it .... if I want to discard something I will Smile | :)

In vino veritas

modified 5-Apr-19 9:40am.

John R. Shaw5-Apr-19 11:00

John R. Shaw

5-Apr-19 11:00

LOL

I think someone got tired of programmers who do not follow the good practice of checking returned values and decided to provide a standard way to warn them. The problem with that is those same programmers also tend to ignore warnings. Wink | ;)

Rajesh R Subramanian12-Apr-19 0:34

Rajesh R Subramanian

12-Apr-19 0:34

John R. Shaw wrote:
... same programmers also tend to ignore warnings.

But, but, if it compiles, I'd then not need to worry about the warnings, right?!

triplet or what ?

Vaclav_31-Mar-19 5:55

Vaclav_

31-Mar-19 5:55

Hope can post this question on this forum.

I was under believe that triplet means three in format:
CPU-company-OS

So why is config.guess adding -gnu as fourth value?
I do not see any tangible benefits in reporting "gnu".

$ sudo bash config.guess
x86_64-pc-linux-gnu

Cheers

k505431-Mar-19 6:30

31-Mar-19 6:30

to a certain extent its the GNU foundation being picky. the systtem is "linux-gnu" i.e. a linux kernel with GNU infrastructure tools. AFAIK there's no other tools running on a linux system, but that doesn't mean there couldn't be.
why are you using sudo in this situation? it shouldn't be neccessary

Vaclav_31-Mar-19 7:27

Vaclav_

31-Mar-19 7:27

Using sudo because it works.

Now I may have to modify the code to look for "linux-gnu", See no issues doing that.

k505431-Mar-19 8:38

31-Mar-19 8:38

Vaclav_ wrote:
Using sudo because it works.

If it only works with sudo then you've seriously borked your system. If commands work without sudo, then please stop doing that ... eventually you're going to do something like

$ cd /bin
do some more things, time passes, you forget where you are ...
$ sudo rm *  ##whoops ... there goes everything in /bin .. time to recover from backups or re-install, more likely

Don't tell me it will never happen to you ... it just hasn't happened to you yet. Something like that has happened to almost everyone that has had root privileges on unix or unix-line system system. That's partially why login as root is generally discouraged. But if you insist on prefixing almost everything with sudo, you might as well just enable root login on your system and go from there.

Randor 31-Mar-19 8:34

Randor

31-Mar-19 8:34

Hi,

At first glance I thought it had something to do with the old GNU/Linux naming controversy as @k5054 is implying. But I think the last field is actually the C library used by the operating system. I just checked on some of my Qualcomm ARM devices and the fourth field is uClibc.

Best Wishes,
-David Delaune

k505431-Mar-19 8:45

31-Mar-19 8:45

so that would be armXX-unknown-linux-uClibc, then? Interestingly FreeBSD (i386, 9.1 - old, i know) reports i586-unknown-freebsd9.1. No mention of the the c-library involved.

Randor 31-Mar-19 9:04

Randor

31-Mar-19 9:04

Well,

Rather than speculate about what it could be let's just take a direct look at the latest source code;

gcc/config.guess at master · gcc-mirror/gcc · GitHub

case "$UNAME_SYSTEM" in Linux|GNU|GNU/*)
    # If the system lacks a compiler, then just pick glibc.
    # We could probably try harder.
    LIBC=gnu
    #if defined(__UCLIBC__)
    LIBC=uclibc
    #elif defined(__dietlibc__)
    LIBC=dietlibc
    #else
    LIBC=gnu
    #endif
    EOF
    eval "`$CC_FOR_BUILD -E "$dummy.c" 2>/dev/null | grep '^LIBC' | sed 's, ,,g'`"
    # If ldd exists, use it to detect musl libc.
    if command -v ldd >/dev/null && \
        ldd --version 2>&1 | grep -q ^musl
    then
        LIBC=musl
    fi
    ;;
esac

So looks like the possible values are:

gnu
uclibc
dietlibc
musl

Best Wishes,
-David Delaune

Vaclav_31-Mar-19 16:20

Vaclav_

31-Mar-19 16:20

It is actually even more convoluted.
If I want to cross-compile - in Eclipse IDE I add "prefix "arm-linux-gnueabi- which translates to "arm-linux-gnueabi-gcc"

That is the actual name of executable cross compiler, not a meaningless triplet.

However , config.guess on Raspberry Pi (armv7l) is "armv7l-unknown-linux-gnueabihf".
That implies that the last member is not just some political peeve.

All this makes the package I am working on useless ( straight from the box) since it does not use the latest version of config.guess in the first place and has "configure" script with no code provisions to analyze the overgrown triplet.

Randor 31-Mar-19 17:21

Randor

31-Mar-19 17:21

Yeah,

Vaclav_ wrote:
If I want to cross-compile - in Eclipse IDE I add "prefix "arm-linux-gnueabi- which translates to "arm-linux-gnueabi-gcc"

Reviewing the config.guess source I can see that it will append libc characteristics.

Vaclav_ wrote:
All this makes the package I am working on useless ( straight from the box) since it does not use the latest version of config.guess in the first place and has "configure" script with no code provisions to analyze the overgrown triplet.

Look, as you probably already know I am primarily a Microsoft technology guy. I left the Linux world over 15 years ago so I am extremely rusty on modern Linux. But I am currently using Linux again on a daily basis and cross-compiling my project so that it will run in Windows, Linux and hopefully any other embedded environments.

Maybe you should try my technique for cross compiling... I just build a quick chroot environment and custom-build the toolchain for whatever platform I am targeting. That's the way I've always cross compiled... but maybe I am using outdated techniques.

Best Wishes,
-David Delaune

[EDIT]

If you are feeling lazy I found an automated script on GitHub that should get you a working build environment for targeting the Raspberry Pi.

Creates a chroot environment for creating custom Raspbian images and cross-compiling programs for the Raspberry Pi.