TRPL: const and static

[steveklabnik]

r? @alexcrichton

[apasel422]

This should be a regular apostrophe, not a curly one.

[steveklabnik]

i thought screwed one up, and I even double checked, sigh

[alexcrichton]

This sentence kinda contradicts the above claim that they can store any reference with a 'static lifetime I think. I think that this may want to remove the reference to the 'static lifetime and focus on the constant/inlining aspect.

Constants should be used for values that do not need an address in memory (as you've said), but they are allowed to reference addresses (e.g. the &'static str case), but that's mostly an implementation detail that the contents of the reference are stored in static memory somewhere.

[steveklabnik]

Fair enough!

[alexcrichton]

Some other topics this may wish to cover (but maybe not here?)

• Guidelines on when to choose a static or a const (it's almost always const)
• Clarification that the type of a static must be Sync.
• Clarification that initialization of both types must be a constant expression (e.g. no CTFE or anything)

[steveklabnik]

@alexcrichton so, when i was writing this, i almost thought they should be together. your additional topics pushed me over the edge. So now, they're combined together, and I tried to add some of the stuff you suggsted. Do you have any better justification for why const over static?

[alexcrichton]

It's pretty rare that you actually want a memory location associated with your constant, and using a const allows for optimizations like constant propagation not only in your crate but downstream crates. A const can be thought of as a #define in C where it has metadata overhead (e.g. it's in a header file or in Rust's metadata) but it has no runtime overhead (as the definition is inlined everywhere).

So I guess the best analogy would be "should I use a #define or a static in C" as it's largely the same question as whether you should use a const or a static in Rust.

Does that help?

[steveklabnik]

@alexcrichton updated!

[alexcrichton]

Does this % render ok? I think we specially handle the first line with a % but not other lines?

[huonw]

I suspect it's a typo for #?

[steveklabnik]

it is

[alexcrichton]

r=me with a minor rendering nit

Thanks @steveklabnik!

[steveklabnik]

@bors: r=alexcrichton rollup

[bors]

📌 Commit a04f6c5 has been approved by alexcrichton

[bors]

⌛ Testing commit a04f6c5 with merge 50cd2e8...

[reem]

I think this would read easier as just:

Constants in Rust have no fixed address in memory because they're effectively inlined to each place they are used. For this reason, references to the same constant are not guaranteed to refer to the same memory address.

Particularly just eliminating the bit about how long constants live, because it's confusing compared to the description of statics.

[bors]

☀️ Test successful - auto-linux-32-nopt-t, auto-linux-32-opt, auto-linux-64-nopt-t, auto-linux-64-opt, auto-linux-64-x-android-t, auto-mac-32-opt, auto-mac-64-nopt-t, auto-mac-64-opt, auto-win-32-nopt-t, auto-win-32-opt, auto-win-64-nopt-t, auto-win-64-opt