CODEWITHSUNDEEP
cmemoryc23
CPlus
CPlus

Reputation: 4848

Is there any difference between the C23 _BitInt() and a non-bit-precise integer of the same width?

The upcoming C23 Standard adds a keyword _BitInt() which can be used, as I understand, to define an integer with a specific number of bits. However I could not find much information with regards to the in-memory representation of types declared this way, and any behavior that relates to their in-memory representation such as their size or alignment.

As such, is there any difference in terms of behavior, representation, or alignment requirements between _BitInt() types and 'real' integer types of the same bit width? For example, between _BitInt(32) and int32_t or int_least32_t? And is it well-defined to type-pun between them?

Upvotes: 4

Views: 2406

Answers (1)

user2357112
user2357112

Reputation: 281748

One important behavior difference is that _BitInt types are exempt from the integer promotions. Adding two int16_t values produces an int, while adding two _BitInt(16) values produces a _BitInt(16). Multiplying two unsigned _BitInt(16) values produces an unsigned _BitInt(16), while multiplying two uint16_t values on a platform with 32-bit ints produces an int (and possible signed overflow).

Upvotes: 3

Related Questions