Detect use of `block.timestamp` and `block.number` in Solidity tests

[PaulRBerg]

Problem

block.timestamp is assumed to be constant when --ir-minimum (via IR) is enabled, and this leads to false negative failed tests when running Solidity tests that use the vm.warp cheat in Foundry.

This can easily create hard-to-debug circumstances - tests that normally pass start failing, and debugging via IR-enabled code is not easy.

Solution

Detect the use of block.timestamp in the users' tests, and throw a bespoke warning about it. Recommend using vm.getBlockTimestamp() instead.

Ditto for block.number and vm.roll and vm.getBlockNumber().

Note

The rule should be turned off by default so that users can turn it on in their test/.solhint.json file.

Related

foundry-rs/foundry#11598 (comment)

[dbale-altoros]

so, to understand

you are proposing a rule , to analyze .sol files with
block.timestamp and suggest vm.getBlockTimestamp
block.number and suggest vm.getBlockNumber
vm.roll and suggest vm.getBlockNumber

?

to narrow it down, this should be done in functions names starting with test or test_
or the file containing the *.t.sol
or inside a test folder

if not, there will be a lot of false positives

[PaulRBerg]

to analyze .sol files with block.timestamp and suggest vm.getBlockTimestamp block.number and suggest vm.getBlockNumber vm.roll and suggest vm.getBlockNumber

Yes, but we can nevermind vm.roll - let's look only at block.timestamp and block.number.

or inside a test folder

Best option would be to look into the user-defined test directory in foundry.toml. For example, in our case, we use tests (plural), not just test.

but if Solhint attempts to look at all *.t.sol files anyway, then that's fine.

[dbale-altoros]

if you can check this new rule|
#719
foundry-no-block-time-number

CONFIG:
foundry-no-block-time-number': ['warn', ['test', 'tests']]
That is the default

Can also be used like this
'foundry-no-block-time-number':'warn'
Will search in default directories (recursively)

Config can be edited to search in any folder

[PaulRBerg]

Looks great!! thanks