Hacker Newsnew | past | comments | ask | show | jobs | submitlogin
C++ value category cheat-sheet [pdf] (github.com/jeaye)
138 points by Jeaye on March 18, 2017 | hide | past | favorite | 39 comments


Softer entry to the whole concept of lvalues/rvalues in general:

http://eli.thegreenplace.net/2011/12/15/understanding-lvalue...


And of course there's Strachey's Fundamental Concepts in Programming Languages (1967), which introduced the LValue/RValue distinction. It is very readable and relevant today.

http://www.cs.cmu.edu/~crary/819-f09/Strachey67.pdf


It is great, but regarding the format I would suggest some options to PDF:

a) markdown with embedded graphviz.

ex: https://github.com/TLmaK0/gravizo

b) a jupyter notebook (which also uses markdown blocks). jupyter can embed C++, so you can have interactive blocks that you can evaluate and play with.

ex: https://github.com/minrk/clingkernel/blob/master/cling.ipynb

Both seem more performant than in-browser PDF rendering.


I looked for a number of options and I didn't find anything which could provide me with automatic syntax highlighting, non-html layouts, and quick iteration. That doesn't mean that such a thing doesn't exist!

Thanks for the suggestions.


Jupyter notebooks are probably what you are looking for. I highly recommend them. Embedded REPL makes iteration very fast and results can be shared directly through a URL, exported to various formats or shared in websites that support it such as github.


Question: it is ok to initialize a string_view from a string literal, right?

e.g.

  std::string_view sv{ "lvalue" };
or

  auto sv{ "lvalue"sv };


Yeah. String literals are lvalues and they live for the lifetime of the program. Both of those snippets would be fine.


Yes, but never use {} initialization with auto.


Never give advice without telling why, because it's very likely that people will forget it if they don't understand your advice.


Ok.

  auto i{42}; // std::initializer_list<int>


No longer applies with c++17



While you are correct, many companies are now just transitioning into C++11, so knowing this will be relevant for the next decade at least.


But why is that bad?


Because `i` isn't an int, it's an initializer_list, and you almost never want to actually create an initializer_list instead of an int? Seems pretty understandable to me.


It's unexpected. Nobody ever* wanted to assign an initializer list to a named variable.

[*] pretty much


I think the biggest issue in C++ is that you don't have to understand this whole mess to be productive in it. This is also perhaps its greatest strength. Where you have Rust which won't even compile a basic toy program without demonstrating a full understanding of its Ownership model.


It isn't a strength to allow ill-defined programs to silently corrupt your process.

If you don't understand what you are writing, either the compiler (rust-like) or the runtime (script-like) should complain.

C++ usually does neither. I think that's undesirable. Why it does neither makes sense, but I wouldn't call it a strength.


Which is it? C++ programmers are smart enough to not need the features that Rust provides? Or C++ programmers aren't smart enough to learn basic concepts that they should already know considering their compiler doesn't help them write correct code?


Are there cheat sheets like this for other languages too?


Most of them aren't as insane.



I'd love it if someone could shed some light on my question here:

https://www.reddit.com/r/cpp/comments/5zzurr/c_value_categor...

    int const &f(int const &a) { return a; }
    int r = f(3);
Is the value of r implementation-defined according to the standard?


This might answer that https://herbsutter.com/2008/01/01/gotw-88-a-candidate-for-th...

I think this is similar to the std::max/std::min stuff


No, that post refers to the lifetime of temporaries held by local reference variables. My question is about temporaries created to pass reference parameters to a function. Is the lifetime of that temporary the BODY of the function? The full expression that calls the function? The enclosing scope of that full expression? (for the record, VS2017 does the latter)


I don't see a problem here. r receives a copy of a, not a reference to anything.

Or have I suddenly gone blind by looking at C++ code one too many times? :)


If the lifetime of the temporary holding the value 3 can (implementation-defined) be limited to the body of the function, the assignment to r happens when the temporary can (implementation-defined) be destroyed already.


Ah, now I see what you mean.

This is not a problem. The temporary ceases to exist after the function call expression. Copying the return value has to be done before that, otherwise you could never return the value of any local variable.


That's what I had always thought would happen; it makes sense. However, if I'm reading things correctly, the posted article disagrees, and this link also disagrees:

http://www.open-std.org/jtc1/sc22/wg21/docs/papers/2016/p013...

"It is implementation-defined whether the lifetime of a parameter ends when the function in which it is defined returns or at the end of the enclosing full-expression."

Hence why I asked for some substantive clarification or solid interpretation of the standard. "Common sense" doesn't cut it with C++. :)


I'm not sure where you see a disagreement with section 5.2.2/4 of the standard:

"It is implementation-defined whether the lifetime of a parameter ends when the function in which it is defined returns or at the end of the enclosing full-expression."

Regardless of where exactly the lifetime of the temporary ends (end of function or end of expression), your example would be correct and not UB.

I don't think it is in dispute whether or not the return expression inside the function is evaluated before the lifetime of the parameter ends, but that is when the copy of a into r is made.

Where do you see a disagreement with the posted article?


If the lifetime of the temporary ended with the function body, the assignment (which reads the temporary via the returned reference) happens afterwards and that's UB. Essentially,

    int const &r = f(3);
and

    int r = f(3);
would both be UB under that implementation-defined case.

The posted article, second row of "common mistakes", describes returning a const reference parameter, when called with a prvalue, as blanket UB.

Finally someone clarified: https://timsong-cpp.github.io/cppwp/class.temporary#6.1

"6.1 A temporary object bound to a reference parameter in a function call persists until the completion of the full-expression containing the call."

So, while the parameter's (the reference "int const &a") lifetime is implementation-defined, the temporary to which that parameter refers to is the full expression of the call. Sanity is restored.


Just to help bring things home:

  const int& f(int parameter) { return parameter; }
  const int i = f(42);
Is strictly undefined behavior under the simpler C++03 rules/wording of 5.2.2/4: "The lifetime of a parameter ends when the function in which it is defined returns.")

P0135R0 would loosen the requirements on parameter's lifetime as a means of enabling guaranteed copy elision, in such a way that an implementation could technically decide to make this to have defined behavior (not that I suggest relying on it!)


Thanks for digging into this. I'll update that example to be more clear about keeping an lvalue reference to the return value.


You're right, good point.


A comment on the document's design: The low contrast of light gray text on yellow background makes it difficult to read. Black text on white, or dark gray text on white is perfectly acceptable.

Another drawback of colored backgrounds is that readers wanting to print out the pdf will consume ink/toner to render the non-white background.


I'm pretty sure the author was trying to use the Solarized color scheme: http://ethanschoonover.com/solarized




That site is great but it leaves out something else that is very important. Color blind people often have trouble with seeing colors that are "near" to each other in color-space.

I'm not a PhD in Colorology but choosing higher contrast colors completely eliminates that problem I think. It's an easy fix that opens up many new markets to your software.




Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: