- widely used by C++ programmers
- int* p, q wrongly implies that both p and q are pointers (leading to a preference for declaring this on two lines, which also improves readability when there are assignments, and makes it easier to quickly cut/paste or comment specific lines/variables)
- int* p visually separates the type from the identifier
- *p then unambiguously indicates a dereference (assuming you put spaces around your binary operator* ala 2 * 3 )
- in C++ . &x is clearly taking an address while . & x must be declaring a reference variable, and . & . is the bitwise-AND operator
- widely used by C programmers
- int *p, q clearly reflects p being a pointer and q not being.
- int *p visually confuses the type with the identifier
- visually indistinguishable from a pointer dereference (for better or worse)
Similarly for types appearing in function declarations.
int* f(), g(); // declares int g(); int *h(), (*i)(); // i is pointer to function returning int int *const*m(), n(); // m returns pointer to (const-pointer to int) // n returns int . but at least function arguments can't get so hairy - the type specification starts afresh after each comma separators.
Summarily, int *p is better if your coding style / code-base utilises multiple declarations on a single line of source code, otherwise int* p offers a clearer separation of type and the following identifier.
For all that, people's preferences are largely based on what they're used to.
