Printable View
I just red this stackoverflow post and still don't get it. Aside from serializing to file, when would an int not work and a uint32 would be better? Same question for uint16 and uint8. I know uint32 means the size reserved for the variable is gauranteed to be 32 bits but 1) why would that matter? 2)since the range is defined of an int wouldn't that imply the size is always the same anyways, i.e. log_2(2^32)=2
The minimum range of int is defined, but implementations are free to exceed it. In fact, mathematically, we only need 16 bits to cover the minimum range of an int, not 32.Quote:
Originally Posted by c_weed
AFAIK, the standard only mentions it has to be larger than or same as a short and smaller than or same as a long.
When you need exactly 32 bits...say for storing 32 boolean flags, it should be a good idea to use the exact type.
Ok. No ones answered when they should be used. For example if it doesn't matter, like your getting user input, should int or uint16 be used?
It is true that the C++ standard does not define the minimum limits but rather the relative sizes. However, it makes references to the C standard on this point, which does define the minimum limits.Quote:
Originally Posted by manasij7479
manasij7479 gave the example of boolean flags, though in C++ there's std::bitset for that. There are certain algorithms that might have such a requirement of number of bits, e.g., the kind that deals with cryptography. In the example of serialization that was cited to you earlier, portability might be the concern, i.e., a number serialized on one platform should be unserialized without loss of data on another platform.Quote:
Originally Posted by c_weed
I would probably choose int because uint16_t is not guaranteed to be available, but it really depends on your requirements. Since you say "it doesn't matter", then you have no requirements, so generate a random number to pick your integer type (or use std::string instead).Quote:
Originally Posted by c_weed
You could use it to store rgba values for 32 bit color depth, or bit flags, or store any unsigned value that wouldn't fit in a 16 bit unsigned integer type. If you're getting user input, you just pick a type that will hold the input (also keeping in mind all the operations you will be doing with that variable).