Resolve confusion of #[test] vs. #[ink::test]

[cmichi]

Escalated from use-ink/cargo-contract#190.

Problem

There is often confusion when to use #[test] vs. #[ink::test].

The implication of #[ink::test] is that the test annotated with it is run in a emulated chain environment (we call this "off-chain testing"). This comes with a bit of overhead for spawning this emulation. This emulation is not needed for every test, only for the ones where e.g. chain storage or the env() API is involved.

For users starting out with ink! it can be hard to know when it's necessary to have an emulated chain, which can result in strange errors if #[test] is used but an emulation is necessary.

Proposals

1. Just replace #[test] with #[ink::test] for every single test and live with the slight performance overhead.
2. Improve naming and split attribute like this:

   #[test]
   #[ink::emulate_chain]
   fn foo_test() { ... }
   

My opinion

My opinion is that we will always have this issue as long as we use different attributes for tests which require an emulated chain and ones which don't. I would just replace #[test] with #[ink::test] in all examples and public documentation and trade a slight decrease in performance for an improved UX with less friction.

[hychen]

I prefer proposal#1. As a developer who builds something on ink!, I'll assume I should always use [ink!...] macros provided by ink! everywhere as minimum build blocks of my smart contracts. To figure out when to use #[test], when to use #[ink:test] is kind of an abstract leak to me.

[cmichi]

All necessary repos have been adapted, closing this issue.