问题
I have some ancient memories of writing C code like:
long value = 0;
in the bad old Win16 days and ending up with value being only half-initialized: i.e. the lower 16 bits were 0 and the upper 16 bits were whatever random bits were at that location in memory. As such, I became conditioned to write:
long value = 0L;
Is this still required in this day-and-age under C99 and/or C++? I know that ILP32 specifies that int and long are both 32-bit, but suppose we're using, say, LP64 where ints are 32-bits and longs are 64-bits. Are the suffixes required or will the modern forms of C and C++ implicitly extend literals to the length of the variable they're being assigned to?
How about unsigned values? I.e. is this required?
unsigned long value = 0UL;
回答1:
They are not required in the examples you gave. However, they may be needed in some somewhat recondite circumstances. For example, the following may produce different values:
sizeof(1)
sizeof(1L)
回答2:
No this should not be required any more. The behavior you are describing, if visible in the program and not just the debugger, is a bug.
回答3:
Since you are presuming a 32-bit compiler, you may not need L as you did when using a 16-bit compiler, but you could need LL, e.g.:
long long ll;
ll = 1<<32;
gcc -std=c99 warns "left shift count >= width of type." Instead, you'd want:
ll = 1LL<<32;
See also https://stackoverflow.com/a/8108658/782738, for C++11 and C99 standard references.
回答4:
But this notation is required for wchar_t strings, like
L"I am a wchar_t string."
and for long long integers, too...
I guess this is something that should be removed, it isn't useful (the type is declared with its name and not with some suffix), but still supported in C++11.
Hm.
来源：https://stackoverflow.com/questions/975942/are-long-suffix-and-unsigned-suffix-needed-when-declaring-long-literals-in-c