问题
When I create models in core data, I'm always a little perplexed by which integer type I should choose—16, 32, 64. I'm almost always needing something for a simple, basic number: a count of people in the household, for my present case. Probably going to be a number between 1-20. Or, I have an incrementing case id number in another instance...can't imagine that going further than few hundred people.
And here's the deal...It's clear that true computer science folk think of numbers differently, taking into account factors like the architecture that's going to be processing the numbers, the space required to process and store the data, backwards compatibility, future proofing, etc. When I think of numbers, I basically think of how large a value is being represented. So when I get to that point of my process when I have to choose between three types of integers, I basically say to myself, "Well, this is going to be a small number, let's just use the Int 16 option...", or "Shoot, I could end up with a really big number here so let's use the Int 64 choice." Basically, I pick these data types with the same sort of logic I use when ordering fries...if I'm really hungry I go for the large, if I'm feeling a big guilty I'll just get the small.
I'm learning enough to know that I'm not thinking about this in the right terms, but I don't really know why, and I don't know the appropriate way to choose the best option. What factors should I really be considering...what's the most important criteria for selecting between Int 16, Int 32, and Int 64?
回答1:
It doesn't matter much.
Assuming you're using a SQLite persistent store, the three integer types are all represented as SQLite INTEGER fields (same for Core Data's "Boolean" type). And in SQLite field type is purely advisory anyway, so even that doesn't mean much. Therefore: it makes literally no difference in terms of storage space. SQLite will optimize itself based on how big the integer values are, and larger int types at the Core Data level will have no effect.
For memory usage, it might have a small impact. If you use a 64 bit int instead of a 16 bit, you're requesting more bits than you need. But unless you have extremely large data sets, it's unlikely that you'll ever have a reason to care.
My usual rule then, is to use Integer 64 for any integral value.
回答2:
You do it correctly: The size of the maximum expected number. But do not care to much about it. Even if you choose a integer "too big", is doesn't fail, but simply uses more memory than needed. On disk.
I think that different sized integers are an anachronism.
来源：https://stackoverflow.com/questions/32892954/choosing-an-integer-type-in-core-data