stacksgov
SIP - Contract Documentation Standard
Creating this SIP as a draft to collect ideas and feedbacks. What's our take on standardizing contract documentation? If we were sharing a standard (a la doxygen or something), like:
;; @contract Counter contract
;; @version 1
(define-fungible-token tok ...)
;; @desc increment the internal counter
;; @param a; value of the increment
;; @post stx; will be transferred to contract
;; @post tok; will be minted for new owner
(define-public (increment (a uint))
...)
Then I think that the overall experience for users interacting with contracts could be enhanced, by explaining what will be done with args and post conditions.
Ideally, if we can converge on a convention, then the stacks-api could handle the mapping, and expose a documented ABI, that wallet can consume.
Thoughts?
Agree with this. Arkadiko is following this convention on all their public functions.
The sip should contain a specification about documenting a minified contract, in particular about adding a reference to a fully documented version of the contract. That would imply a standardization of minification. Is that out of scope?
I agree this would be very valuable. I'd love to see something included that would allow wallets and explorers to show details about errors as well. For example, something like:
;; @err Balance too low
(define-constant balance-too-low-err (err u67))
This could then be used to show "Error: Balance too low" instead of just showing "(err u67)".
Instead of storing this data on-chain, a
@error-uri "https://..."
would be more efficient, and would better support localization.
I do like the idea of having the contract point off-chain for detailed documentation, but what would be the argument for having some structured on-chain documentation and some off-chain? Is that because we could consider the description, parameter explanation, and post conditions vital to the security, but the error message as just a convenience?
If the documentation is stored to a link off-chain, the documentation can change independently of the contract. This has positive and negative aspects. On one hand, it gives you more flexibility but on the other it also allows a malicious user to alter the documentation with misguided information without touching the contract. A way to solve this can be to store in chain a hash of the documentation, to guarantee that the documentation is valid for that contract. Not sure if you need to go down that road. Just thinking out-loud here about possible implications...
If the documentation is stored to a link off-chain, the documentation can change independently of the contract.
The hash(es) of the documentation could be stored on-chain if this is a concern.
We should consider this topic together with the suggestion in https://github.com/blockstack/stacks-blockchain/issues/2926 which discusses storing a minified version of the contract. If we decide to do that, then it would reduce the cost of including significant amounts of comments directly in the contract.
We'd be very interested in leveraging the data that conforms to such a standard in the Hiro Wallet during transaction signing to better present and explain the contract being signed. Currently, the contract name and arguments alone can be quite cryptic for non-technical users.
We should consider this topic together with the suggestion in stacks-network/stacks-blockchain#2926 which discusses storing a minified version of the contract.
To be clear, stacks-network/stacks-blockchain#2926 refers to compression, not minification. I want to emphasize that the compression algorithm will not attempt to parse the contract code body.
I suggest that we use a specific notation for doc comments (such as ;;; instead of ;;)
It would allow some doc specific features such as scaffolding of docs section / auto-completion and would make syntax highlighting easier.
Just like JSDoc uses /** */ instead of /* */
We had a discussion about ;; and ;;; where the latter is used for docs that can be removed before deployment for cheaper costs.
I think I could find some use case where it would be useful to have it on mainnet. But yes, if deployment costs are an important consideration, we could have an option to remove docs comment on deploy
I am all for a reasonable documentation on-chain.
How to represent
- on chain docs, including docs for doc-gen
- off-chain docs, including docs not for doc-gen
@friedger I see, I didn't consider "off-chain docs, including docs not for doc-gen" Not sure if it's in the scope of this of this SIP but definitely something to take into consideration.
Anyway, I'm in favor of using a special comment notation for on chain docs, including docs for doc-gen
Ludo suggested keywords starting with @ for docs-gen.
Combined with ;;; we could cover all cases.
yeah that could work but I'm not sure it would be perfect. I think it would be easier to have the discussion on a PR comments or on Discord rather than in the PR main thread
Noting some discussion items here from the Clarity WG call:
- We should revitalize this SIP and get it formalized
- With the coming Clarity Wasm, only the built wasm module needs to be loaded on function calls, not the contract source, so the cost of having extra comments should only affect deployment, not every call to the contract.