coverage: both branches lines taken but only half branch coverage

[hellwolf]

Component

Forge

Have you ensured that all of these are up to date?

• Foundry
• Foundryup

What version of Foundry are you on?

forge 0.2.0 (ca67d15 2023-08-02T00:16:51.291393754Z)

What command(s) is the bug in?

forge coverage --hardhat --report lcov

Operating System

Linux

Describe the bug

We found sporadically some cases that the lcov reporting from foundry did not make sense to us, since the branch should in all cases be hit, here are some examples:

• https://app.codecov.io/gh/superfluid-finance/protocol-monorepo/blob/dev/packages%2Fethereum-contracts%2Fcontracts%2Fagreements%2FConstantFlowAgreementV1.sol#L776 (in normal non-branching code path).

• https://app.codecov.io/gh/superfluid-finance/protocol-monorepo/blob/dev/packages%2Fethereum-contracts%2Fcontracts%2Futils%2FBatchLiquidator.sol#L67 (in normal code block)

• https://app.codecov.io/gh/superfluid-finance/protocol-monorepo/blob/dev/packages%2Fethereum-contracts%2Fcontracts%2Futils%2FCFAv1Forwarder.sol#L411 (abi.encodeCall)

• https://app.codecov.io/gh/superfluid-finance/protocol-monorepo/blob/dev/packages%2Fethereum-contracts%2Fcontracts%2Futils%2FIDAv1Forwarder.sol#L21 (normal code paths in a constructor)

... etc.

Is there something we could help with to give more diagnostics information about those?

[dcposch]

False negatives in branch coverage

Here's a clear example. (Ignore the nonsensical else, I added it to confirm this issue:)

forge coverage correctly shows both branches taken (line coverage in both if and else), but incorrectly reports 1/2 branch coverage for the if

using --report debug:

Uncovered for src/DaimoAccountFactory.sol:
- Branch (branch: 0, path: 0) (location: source ID 54, line 39, chars 1597-1685, hits: 0)

[Evalir]

Closing as duplicate, but ack coverage needs more work. Will revisit during next year

[DaniPopes]

What is this a duplicate of? I can't determine if this was ever fixed, so re-opening. CC @grandizzy do you recall if you've fixed a similar issue?

[grandizzy]

@DaniPopes ah, actually yes, there were several fixes re branch report incorrectly, I added unit tests for

#8414
#8433

and several others, I didn't see any issue with branches lately