[Stacks 2.1] Fuzz targets for 2.05 and 2.1 Clarity 1 VM

[jcnelson]

To build confidence in our assessment that the Clarity 1 VM in Stacks 2.1 is truly compatible with the one in Stacks 2.05, we should set up and run fuzzing to check that the typechecked ASTs produced by both VM implementations are identical.

Unfortunately, this appears to be outside the scope of cargo-fuzz because it involves comparing the behaviors of two different git branches. But, we could still create a fuzzer that synthesizes random well-formed Clarity 1 contracts, feeds them into both the 2.05 and 2.1 VM implementations, and checks the resulting ASTs. A suggested implementation tactic would be to set up pipe servers for both 2.05 and 2.1 VMs (e.g. within clarity-cli), and have the fuzzer feed them the same random contract and check the resulting (serialized) annotated ASTs for compatibility.

[whoabuddy]

@moodmosaic this came up in the blockchain meeting and would love to know if you have any input given your work with fuzzing in Clarinet. (also: #3427)

[moodmosaic]

set up pipe servers for both 2.05 and 2.1 VMs (e.g. within clarity-cli), and have the fuzzer feed them the same random contract and check the resulting (serialized) annotated ASTs for compatibility

Can you check/expose the ASTs via clarinet? (Even via a custom build of it.) @obycode @lgalabru — If you can, then roundtrip testing and differential testing may be possible in this scenario, using a currently unreleased version of clarinet.

Over times, while working on this, it randomly helped find and fix a few corner cases:

• hirosystems/clarinet@de04217#r89175880
• Runtime errors when fuzzing a public function taking string-ascii hirosystems/clarinet#496

In its current form, it looks like this: PromptECO/clarity-sequence#4 (comment), though for this ad-hoc scenario it can be something completely different... Given more context, I may have more/different input, however.

[jcnelson]

Temporarily assigning to @fpbgg. Please feel free to re-assign, close, or break down as you see fit.