The intent of that sentence is a lot clearer if you also consider the subsequent text, which expands on the idea quite a bit:
> Computers can only natively store integers, so they need some way of representing decimal numbers. This representation comes with some degree of inaccuracy. . . Why does this happen? It's actually pretty simple. When you have a base 10 system (like ours), it can only express fractions that use a prime factor of the base. . .
Perhaps a more formally correct way to put it is that integers (and natural numbers) are the only numbers that a computer can manage in a way that behaves reasonably similarly to the corresponding mathematic set. Specifically, as long as you stick to their numeric range, computer ints behave like a group, just like real integers do. But IEEE floats break the definition of a field in every which way, so they're really not a great match for the rationals.
That said, you could represent the rationals as a pair of integers, and that would be better-behaved, and some programming languages do do that. But I'm not aware of an ISA that supports it directly.
I believe that's where "natively" comes in. If you store it in binary, you can operate on the values as numbers. If you do some decimal conversion, it's a blob of data. (Yes, there are BCD instructions but they're massively limited)
Maybe you need to parse "decimal numbers" as "fractional numbers" rather than "expressed using a radix of 10".
Perhaps it would be better phrased as "computers can only store whole numbers, so they need some way to represent other information, including integers, rational numbers and approximations to complex numbers."
> Computers can only natively store integers, so they need some way of representing decimal numbers. This representation comes with some degree of inaccuracy. . . Why does this happen? It's actually pretty simple. When you have a base 10 system (like ours), it can only express fractions that use a prime factor of the base. . .
Perhaps a more formally correct way to put it is that integers (and natural numbers) are the only numbers that a computer can manage in a way that behaves reasonably similarly to the corresponding mathematic set. Specifically, as long as you stick to their numeric range, computer ints behave like a group, just like real integers do. But IEEE floats break the definition of a field in every which way, so they're really not a great match for the rationals.
That said, you could represent the rationals as a pair of integers, and that would be better-behaved, and some programming languages do do that. But I'm not aware of an ISA that supports it directly.