Re: TYPEOF

From: jbandela_at_[hidden]
Date: 2001-06-06 13:47:31

> Not likely, since this "solution" requires registering the types 
> explicitly much as the "solution" poted in the files section does.

I think my "solution" has an advantage in that it scale's better. In 
the example code of the article mentioned, changing short to short* 
caused compiler errors. My solution handles most (if not all) 
combinations of those types for example. The following three 
registrations

JRB_TYPE0(int)
JRB_TYPE1(allocator)
JRB_TYPE2(vector)

could handle
int*, int**, const int*, int * const, volatile const int,
vector<int>, vector<const int*>, vector<vector<int*>&>, and so on

  
> This is not a library replacement for a language typeof().

No it is not, but it can be useful. In C++ programming, though there 
are often a lot of types used, they are composed of much 
fewer "basic" types.  For example, Typelists.

Typelist<int,Typelist<const 
int,<Typelist<int*,Typelist<int**,NullType> > > > > (I may not have 
got all the angles braces)

Has 9 types (at least). However it is compolsed of only 3 entities
1. a built in type - int
2. a non-template class NullType
3. A template named Typelist that takes 2 type parameters

Those three entities can describe any combination Typelist, NullType, 
int, and any pointers, pointers to pointers, const, volatile, etc