SDK release and backward compatibility

[ilblackdragon]

Currently when something changes on nearcore runtime side we may have a corresponding change in sdk. But that new sdk doesn't work with stable version of nearcore.

We need to either make sdk work in backward compatible way or not release new sdk until we have code change required made into stable

[MaksymZavershynskyi]

In the Middleware team we have discussed the release scheme that would prevent such confusion. When adding new host functions there is no simple way to make it backward compatible because of how Wasm compilation works and how pwasm-utils perform static checks.

Specifically for near-sdk-rs, near-sdk-as we will do the following we will be releasing the version with -beta suffix to emphasize that it works only on betanet before promoting this to a release without suffix. This scheme might also be adopted by other tools like: near-api-js.

Overall, we have decided to perform the following items for near-sdk-rs, borsh, runtime crates:

• Release 1.0 version;
• Start using change log like here: https://github.com/near/near-sdk-rs/blob/master/CHANGELOG.md If crate is the single crate in the repo (not applied to the runtime) then also use github releases: https://github.com/near/near-sdk-rs/releases
• Coordinate upcoming API changes using changelog between each other, Applayer, and DevX teams.

Overall, I am going to bring up the issues of coordinating crate releases during our upcoming eng sync. I think we need to cleanup versioning of nearcore crates, including adding changelogs, and having cross-team coordination.

I am closing this issue, since we have discussed the solution.