Is there a way to snapshot storage between tests instead of redeploying every time?

[sigittt1998]

My test suite deploys a bunch of contracts in setUp() and each test takes ~2 seconds because of it. The contracts are the same across all tests — only the test actions differ.

I thought vm.snapshot() and vm.revertTo() would help but they seem to work within a single test, not across tests.

Is there a way to deploy once and reuse that state across the whole test file? Something like a shared fixture that all tests fork from? My suite has 40+ tests and it takes over a minute to run which is brutal for iteration.

Using forge 1.1.x if that matters.

[yudin-s]

Not in the “deploy once for the whole test file” sense. In Forge’s normal Solidity test model, setUp() runs before each test, so each test gets its own setup phase. The state snapshot cheatcodes are for snapshotting and restoring EVM state inside a test execution context; they are not a cross-test fixture cache that survives into the next test.

So this works for branching inside one test:

function test_manyBranches() public {
    uint256 snap = vm.snapshotState();

    // branch A
    // ...

    vm.revertToState(snap);

    // branch B from the same setup state
    // ...
}

but it is not a way to do:

deploy once -> run 40 independent test functions from that deployed state

If the expensive part is deployment, the usual Foundry options are:

• Keep unit tests local and make setUp() cheaper: deploy only the contracts required for that test contract, split large fixtures across multiple test contracts, and avoid broad integration setup in every file.
• For live/protocol state, use fork tests pinned to a block; Foundry caches fork RPC data, which helps repeated runs, but it does not cache your Solidity setUp() deployments as a shared mutable fixture.
• If several assertions truly share the exact same fixture and do not need isolation, group related branches inside one test with snapshotState() / revertToState(). That trades test granularity for iteration speed, so I would use it only for tightly related scenarios.
• Run a narrowed subset while iterating, e.g. --match-contract / --match-test, and keep the full 40-test suite for CI or pre-push.

In short: vm.snapshotState() is useful within a test; setUp() is still per test. If this answers it, please mark it as answered so future readers do not go hunting for a hidden cross-test snapshot feature.

[sigittt1998]

setUp() with snapshot + revertTo in each test is exactly what I needed, cuts suite time in half. thanks