Reputation: 6882
I try to compute a constant depending on a type and some other value with some template metaprogramming.
template <typename t, uint8_t number_of_bits> struct bin_size {};
template <>
struct bin_size<uint8_t, uint8_t number_of_bits> {
const uint8_t upper_bound = 255;
};
template <>
struct bin_size<int32_t, uint8_t number_of_bits> {
const uint8_t upper_bound = 60 * number_of_bits * 10;
};
However the compiler (arm-none-eabi-g++ (GNU Tools for ARM Embedded Processors (Arduino build)) 4.8.3 20140228 (release) [ARM/embedded-4_8-branch revision 208322] ) complains with the following errors.
test.cpp:287:52: error: template argument 2 is invalid
struct bin_size<uint8_t, uint8_t number_of_bits> {
^
test.cpp:292:52: error: template argument 2 is invalid
struct bin_size<int32_t, uint8_t number_of_bits> {
^
Error compiling.
Without the number_of_bits feature everything works out as it should. But I can not figure out how to specialize on the typename but not on the number of bits. How can this be done?
Upvotes: 1
Views: 112
Reputation: 96810
Make number_of_bits
a template argument:
template <uint8_t number_of_bits>
struct bin_size<uint8_t, number_of_bits> {
const uint8_t upper_bound = 255;
};
template <uint8_t number_of_bits>
struct bin_size<int32_t, number_of_bits> {
const uint8 upper_bound = 60 * number_of_bits * 10;
};
Upvotes: 3
Reputation: 3389
Simply add a template parameter expecting a number and use its name in your specialization :
template <uint8_t number_of_bits>
struct bin_size<uint8_t, number_of_bits> {
const uint8_t upper_bound = 255;
};
template <uint8_t number_of_bits>
struct bin_size<int32_t, number_of_bits> {
const uint8_t upper_bound = 60 * number_of_bits * 10; // You forgot "_t" here.
};
By doing so, the specialization is partial and still depends on something (number_of_bits
in your case)
Here is an example : https://ideone.com/fvTa0O
Upvotes: 3
Reputation: 217275
The (partial) specialization should be something like:
template <uint8_t number_of_bits>
struct bin_size<uint8_t, number_of_bits> {
const uint8_t upper_bound = 255;
};
Upvotes: 3