Re: [eigen] Value-initializing a Matrix |
[ Thread Index | Date Index | More lists.tuxfamily.org/eigen Archives ]
I hope Eigen stays compatible with non C++11 compilers.
I dont get your code?:Do you mean this:struct foo {
foo() : v{}
private:
Eigen::Vector2d v;
};
or
struct foo {
foo() : v()
private:
Eigen::Vector2d v;
};I think Eigen should support value initialization ( although its tricky business and I am not quite sure about all the different initialization, “default”, “value”, “zero”)I think Eigen3 should (at least in the next release ) include the c++11 conformance with value initialization such that v{} get initialized properly (that means to zero)and any default-initialization should result in undetermined values .Anyway how can we distinguish in C++ betweenMatrixXd M; < default-initialization -> uninitializedMatrixXd M{}; < value-initialized -> zero initializedWhat would need to be done to distinguish these two cases?Am 19.04.2015 um 00:44 schrieb Matan Nassau <matan.nassau@xxxxxxxxx>:To be clear, my question stands still whether the code is
struct foo {
foo() : v{} {} // braces
private:
Eigen::Vector2d v();
};
or
struct foo {
foo() : v() {} // parentheses
private:
Eigen::Vector2d v();
};
The question is, what's the value of v after construction? In the current state of affairs it is indeterminate. It can be anything. But one might expect it to be value-initialized just as std::vector or int would be (such that v.size()==0 or v==0 for the std::vector and int respectively) if that's what the type of v were. Yes, I can get this behaviour with the EIGEN_INITIALIZE_MATRICES_BY_ZERO define, but then I can't get the do-nothing performance anymore if that's what I want, so that's not a solution.On Sat, Apr 18, 2015 at 18:26 Matan Nassau <matan.nassau@xxxxxxxxx> wrote:But, there is no initializer list here. There is an =default, yes, but maybe we can do that conditionally?
#if __cplusplus >= 201103L
Matrix()=default;
#else
#ifdef EIGEN_INITIALIZE_MATRICES_BY_ZERO
...
If Eigen supports C++11 initialization semantics it opens the door for more performant code, and for more expressive flexibility in user code. I mean, I can clearly express what I want with
Matrix3d m;
and
Matrix3d m{};
controlling the performance or behaviour in case-by-case fashion, in my code.
Eigen can do this, while still providing the current behaviour for C++03.On Sat, Apr 18, 2015 at 15:47 Gael Guennebaud <gael.guennebaud@xxxxxxxxx> wrote:Hi,
Eigen aims to be compatible with C++03, and initializer-lists are not supported yet. So currently Vector2d v{}; is equivalent to Vector2d v; .
See this bug entry for further discussions on initializer-lists: http://eigen.tuxfamily.org/bz/show_bug.cgi?id=954. They will likely be supported in Eigen 3.3, and we already agreed on making the empty initializer-list initializes to 0.
In the meantime, you can compile with -DEIGEN_INITIALIZE_MATRICES_BY_ZERO, as documented there: http://eigen.tuxfamily.org/dox/TopicPreprocessorDirectives.html#title0.cheers,gaelOn Sat, Apr 18, 2015 at 7:22 PM, Matan Nassau <matan.nassau@xxxxxxxxx> wrote:will create a vector with an undetermined value.Eigen::Vector2d v{};I have a class,struct foo {foo() : v{} {}private:Eigen::Vector2d v;};I just got bitten when I realized the value of v is undetermined.I understand the default constructor of a fixed-size matrix does nothing. In particular,Why is that? Is this for speed?
If I value-initialize an object I expect it to initialize. To motivate, all standard templates and classes behave this way:std::string s{}; // assert(s=="");std::vector v{}; // assert(v.size() == 0);Granted,std::string s; // assert(s=="");but if we want speed here then we can dotemplate<...struct Matrix {Matrix() = default;
// ....This way, we'd get the best of both worlds:Eigen::Vector2d v; // valid, undetermined valueEigen::Vector2d v{}; // assert(v==Eigen::Vector2d::Zero());Matan
Mail converted by MHonArc 2.6.19+ | http://listengine.tuxfamily.org/ |