The Stan Math Library is a C++ template library for automatic differentiation of any order using forward, reverse, and mixed modes. It includes a range of built-in functions for probabilistic modeling, linear algebra, and equation solving.
The various functions in the Math library that deal with sizes of containers return different types than each other:
size() returns size_t (equivalent to unsigned long int)
rows() returns int
cols() returns an Eigen::Index which at least on GCC is equivalent to long int
num_elements() returns int
dims() returns a vector of int
In particular, size_t is problematic because it is an unsized type, which leads to bad semantics when the result is used in a mathematical expression as seen in this forum post
We should probably use the same type for all of these functions. Because int is what is used in the language, that seems like a reasonable choice. If we're worried about internal overflows, we should at least used a signed type like ssize_t/long int, rather than size_t
Description
The various functions in the Math library that deal with sizes of containers return different types than each other:
size()returnssize_t(equivalent tounsigned long int)rows()returnsintcols()returns anEigen::Indexwhich at least on GCC is equivalent tolong intnum_elements()returnsintdims()returns a vector ofintIn particular,
size_tis problematic because it is an unsized type, which leads to bad semantics when the result is used in a mathematical expression as seen in this forum postWe should probably use the same type for all of these functions. Because
intis what is used in the language, that seems like a reasonable choice. If we're worried about internal overflows, we should at least used a signed type likessize_t/long int, rather thansize_tCurrent Version:
v4.9.0