[intro.memory] What is an address?

[TymianekPL]

We never provide the definition of an address yet we refer to it so much...

I would love to work on this one, I have a couple of ways to solve it and am still figuring out which is the best.

PS: We do have a definition of a memory location but it does not answer my question nor does it define an address.

[TymianekPL]

I can open a PR within a couple of hours.

[jensmaurer]

I can open a PR within a couple of hours.

I have no doubts, but I believe adding words that define something as delicate as the term "address", given the memory and object model of C++, is not editorial by far.

See [intro.memory] p1.

Please submit a CWG issue with your proposed definition here: https://github.com/cplusplus/CWG/issues/

[eisenwave]

For what it's worth [intro.memory] p1 seems like a quasi-definition (https://eel.is/c++draft/intro.memory#1.sentence-4):

Every byte has a unique address.

This is essentially defining an "address" as a construct that a byte has one of, considering that it's the first occurrence of the term that resembles a definition. This is similar to how https://eel.is/c++draft/intro.object#def:object defines an "object" in terms of how it's created, not in terms of what it is.

All that is to say, maybe it would be correct to slap italics on "address" in [intro.memory] p1 and we'd have our definition.

[TymianekPL]

All that is to say, maybe it would be correct to slap italics on "address" in [intro.memory] p1 and we'd have our definition.

That was one of my solutions.

Besides, is an address a byte or does a byte have an address?

Every byte has a unique address.
[intro.object]
Unless an object is a bit-field or a subobject of zero size, the address of that object is the address of the first byte it occupies.
[intro.object]

[TymianekPL]

Ah I think I misread the last one, well, still, we need to define what an address is.

[TymianekPL]

I can open a PR within a couple of hours.

I have no doubts, but I believe adding words that define something as delicate as the term "address", given the memory and object model of C++, is not editorial by far.

See [intro.memory] p1.

Please submit a CWG issue with your proposed definition here: https://github.com/cplusplus/CWG/issues/

Alright I'll submit the issue there once I finish a couple of things and come up with wording! Thanks