Add assert_msg(cond, "message") emitting Solidity Error(string)

[cburgdorf]

Introduces assert_msg<const N: usize>(b: bool, message: String<N>) in
the std prelude. On b == false it reverts with the 4-byte selector
0x08c379a0 (keccak256("Error(string)")[..4]) followed by the
ABI-encoded message, matching Solidity's require(cond, "message") /
revert("message").

Also improves const-generic inference so a string literal can satisfy a
generic String<N> parameter without an annotation at the call site:
when a string literal is passed to a parameter of shape String<N> and
the inference key for N appears only in that one parameter, the
checker pins N to the literal's min length. Keys shared across
multiple parameters are left alone so the other arguments (e.g.
[u8; N]) still drive inference for N.

Implements #1348

[cburgdorf]

@codex review

[chatgpt-codex-connector]

💡 Codex Review

Here are some automated review suggestions for this pull request.

Reviewed commit: fb5af0b111

ℹ️ About Codex in GitHub

Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you

• Open a pull request for review
• Mark a draft as ready
• Comment "@codex review".

If Codex has suggestions, it will comment; otherwise it will react with 👍.

Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".

[cburgdorf]

@codex review

[chatgpt-codex-connector]

💡 Codex Review

Here are some automated review suggestions for this pull request.

Reviewed commit: 9151f04914

ℹ️ About Codex in GitHub

Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you

• Open a pull request for review
• Mark a draft as ready
• Comment "@codex review".

If Codex has suggestions, it will comment; otherwise it will react with 👍.

Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".

[cburgdorf]

@codex review

[chatgpt-codex-connector]

💡 Codex Review

Here are some automated review suggestions for this pull request.

Reviewed commit: 404023ca5e

ℹ️ About Codex in GitHub

Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you

• Open a pull request for review
• Mark a draft as ready
• Comment "@codex review".

If Codex has suggestions, it will comment; otherwise it will react with 👍.

Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".

[cburgdorf]

@codex review

[chatgpt-codex-connector]

Codex Review: Didn't find any major issues. You're on a roll.

ℹ️ About Codex in GitHub

Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you

• Open a pull request for review
• Mark a draft as ready
• Comment "@codex review".

If Codex has suggestions, it will comment; otherwise it will react with 👍.

Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".

[sbillig]

Proposal:

Define assert_msg in terms of Text aka DynString:

     use super::abi::Text

     pub fn assert_msg(_ b: own bool, _ message: own Text) {
         if !b {
             let (ptr, len) = encode_calldata(0x08c379a0, message)
             ops::revert(offset: ptr, len: len)
         }
     }

(this might increase gas usage)

and remove the hir changes. There's an issue in hir to be solved, but the current solution isn't the right one. I think we need to modify the type unifier so that it imposes an N >= literal_len constraint for String<N>, which probably isn't a huge change, but could be deferred.

[cburgdorf]

@sbillig ok, made the requested change.

[sbillig]

I think the abi-encoded payload is wrong. I'm rebasing and investigating.

[sbillig]

Here's what the agent says:

Change assert_msg from:

let (ptr, len) = encode_calldata(0x08c379a0, message)

to:

let (ptr, len) = encode_calldata(0x08c379a0, (message,))

with a short comment explaining why.

Why: encode_calldata expects its second argument to represent the full ABI payload after
the selector. Generated msg/custom-error structs already do that. A bare Text/DynString
encodes as its dynamic payload:

len || bytes

but Error(string) has one ABI argument, so it needs the single-argument head first:

offset(0x20) || len || bytes

So for "boom", the correct full revert data is:

08c379a0
0000000000000000000000000000000000000000000000000000000000000020
0000000000000000000000000000000000000000000000000000000000000004
626f6f6d00000000000000000000000000000000000000000000000000000000

[cburgdorf]

Ok, I did what the agent said 🫡

[cburgdorf]

...and added another test that covers this.