What is difference between these two? Which one you would prefer when you need a fixed size array of cons开发者_如何学JAVAtant values?
const boost::array<int, 2> x = {0, 1};
boost::array<const int, 2> y = {0, 1};
Thanks.
The second one will prevent that you copy it to a new non-const array
boost::array<const int, 2> y = {0, 1};
boost::array<int, 2> y1 = y; // error!
Since I would expect that to work, I would probably go with the first option. Passing the second one to templates that expect a boost::array<T, N>
will prevent those templates from modifying their parameter (even if it's a copy). The first one would "just work", since the parameter would have the type boost::array<int, 2>
.
It's really a stylistic difference.
If you try to call assign
on a const array
, the compiler error says there is no matching function. If you do the same with an array<const T>
, it points at the invalid operation inside assign
.
I think const array
expresses intent better, and looks more like the corresponding C-style array declaration. But I wouldn't make an effort to change things, for example in legacy code or inside a template which might generate an array<const T>
.
A const int
and an int
in this context are pretty much the same. There's nothing you can do to an array<int,2>
that you can't do to an array<const int, 2>
. If instead of int
you have some class then there would be a difference. with array<const MyClass, 2>
you would not be able to call non-const methods on the elements of the array.
const array<MyClass, 2>
is stronger in that you cannot modify anything what so ever. You can't call non-const methods of the elements and you can't change the array itself by replacing the elements, say using operator[]
.
精彩评论