Use package name instead of lib name, default to "rlib"

[ascjones]

Closes #927.

I have a vague memory of some reasoning that the package name and lib name were separate. However I cannot remember or imagine what it might be, so LFG.

[ascjones]

CI errors should be fixed in #930

[athei]

I think rlib is the default for libs anyways. Do we even need crate-type in the manifest then?

[HCastano]

I think rlib is the default for libs anyways. Do we even need crate-type in the manifest then?

From here it doesn't necessarily say that it's the default. It just says that the compiler will choose the best option

[HCastano]

Yeah this goes back to at least 2019 🤷

[ascjones]

I think rlib is the default for libs anyways. Do we even need crate-type in the manifest then?

I've tested this with the delegator ink! example, removing the crate-type section from the adder manifest, and the it compiles fine with the same resulting contract size.

[athei]

I think rlib is the default for libs anyways. Do we even need crate-type in the manifest then?

From here it doesn't necessarily say that it's the default. It just says that the compiler will choose the best option

Yes. But the default should always allow depending on it from another crate. We actively prevented this by setting it to only cdylib before this PR. I just tested without any crate type: I can depend on a contract.

I've tested this with the delegator ink! example, removing the crate-type section from the adder manifest, and the it compiles fine with the same resulting contract size.

This was expected because we set it to cdylib during build.

[HCastano]

@ascjones can you add a CHANGELOG entry since this changes the default behaviour of contract dependencies?