Add cargo semver-checks to CI

[vicsn]

Motivation

Proof of concept to showcase: ProvableHQ/snarkVM#2912

You may want to cache the target folder too like we do in that PR, where we get a speedup from from 40 minutes down to 15 minutes.

By default checks are made against the most recent published release. The approach I take in snarkVM is to manually append a --baseline-rev eff84712b if I want to just compare against a recent commit in between releases.

[iamalwaysuncomfortable]

This doesn't seem to work with our manual monomorphization for creating the two wasm binaries, a cursory 15 minutes of trying couldn't seem to get it to work, but we can investigate further later.

[vicsn]

This doesn't seem to work with our manual monomorphization for creating the two wasm binaries, a cursory 15 minutes of trying couldn't seem to get it to work, but we can investigate further later.

Yeah there's no rush on this. I see that it's cargo doc failing, which this relies on. It also failed for a snarkVM subcrate initially which I had to fix. So one added benefit of adding this in CI is that you also get a cargo doc check.

I think this might fix the issue at hand:

diff --git a/wasm/src/types/networks/mod.rs b/wasm/src/types/networks/mod.rs
index 67b6a928..7cb27bf4 100644
--- a/wasm/src/types/networks/mod.rs
+++ b/wasm/src/types/networks/mod.rs
@@ -18,10 +18,10 @@
 mod testnet;
 
 #[cfg(feature = "testnet")]
-pub use testnet::*;
+pub use testnet as snarkvm_network;
 
 #[cfg(feature = "mainnet")]
 mod mainnet;
 
 #[cfg(feature = "mainnet")]
-pub use mainnet::*;
+pub use mainnet as snarkvm_network;

[vicsn]

I relent, on second look, the above is indeed a lot of work, not worth it right now is SDK interface is decently stable

[iamalwaysuncomfortable]

If this can be given attention by a CI engineer, we're happy to have it, otherwise it might be too much of a lift on engineering resources at the moment to get this working against both the JS and rust code.