Hi all -
I have a struct which holds parameter name/value pairs. The incoming
data is not always exactly the same, as it's read from different
sources, so sometimes I might have a parameter named "scale" which is
a uint8_t, and other times "scale" is a uint32_t or int32_t. When I
try to get() the data, I grab it as uint32_t, but it may be stored in
the variant as an int32_t.
We're using get in that case, and this fails if the variant
holds int32_t. On the other hand, since the variant can contain
vector and matrix values, we can't simply use operator = for automatic
casting (compile time errors).
We don't expect to do any non-trivial casts (i.e. matrix to float or
vice-versa). Are there any helper visitors or anything in the
boost::variant library to assist with this, or do we need to write our
own visitor?
Code below.
Thanks,
Brian
struct ParameterDesc {
typedef boost::variant ParamValue;
typedef google::dense_hash_map ParamMap;
typedef ParamMap::iterator ParamMapIter;
typedef ParamMap::const_iterator ParamMapCIter;
std::string type;
std::string name;
ParamMap params;
template <typename T>
bool get(const std::string &name, T &val, const T &defaultValue) const {
ParamMapCIter p_it;
if( (p_it = params.find(name)) == params.end() ) {
val = defaultValue;
return false;
}
val = boost::get<T>(p_it->second);
return true;
}
template <typename T>
void get(const std::string &name, T &val) const {
ParamMapCIter p_it;
if( (p_it = params.find(name)) == params.end() ) {
throw ParamException("couldn't find parameter with field
name: " + name);
}
val = boost::get<T>(p_it->second);
}
ParameterDesc() {
params.set_empty_key("__EMPTY__");
}
};
