Snapshot API saves and restores time deltas.

[scnale]

When running tests, the snapshot API wasn't saving the time adjustments made through the evm_timeIncrease JSON-RPC command. I added the code necessary to save and restore the appropriate state but I'm not sure this is the best way to do it. According to my own tests it seems to be working.

[hiqua]

Is there a reason not to merge these changes?

[madvas]

Please merge, much needed

[benjamincburns]

@scnale sorry it has taken us so long to circle 'round to this PR! Unfortunately, I don't think it's working. I've pushed a (failing) test to this PR which demonstrates how I think it's meant to work. Can you please either update the test to correct my misunderstanding, or update your solution to make the test pass?

[benjamincburns]

I don't know why, but the new test isn't showing in the latest travis build. I promise it's there, and if you run it locally it fails.

[scnale]

Ah, I see. Comparing timestamps won't work because the time adjustment is added to the current timestamp. I think the best way to test this would be to:

• take a snapshot
• call evm_increaseTime with any value other than zero
• restore the snapshot
• call evm_increaseTime with nonzero value k (preferably different from the previous one) and check its return value for equality with k.

Edit: Nevermind, I thought it was comparing timestamps. I will look into it once I get it setup.

[benjamincburns]

@scnale it's been quite a while - no pressure, but do you intend to fix up the test failures? If not, I'll take it on if I have time, or if I don't I'll just close this PR and flag the issue as needshelp.

[ouziel-slama]

Please fix/merge, much needed to write real tests. Thanks.

[benjamincburns]

@ouziel-slama, per comments above, the changes in this PR don't work as intended. Since I haven't heard back from @scnale in quite a while I'll close this to avoid confusion. If you'd like to fork @scnale's changes and submit a new PR yourself, I strongly encourage it!

[PhABC]

@benjamincburns and @scnale the fix works as intended. There is a mistake in the new test case, where revert should revert to snapshots[1]. Line 87 should therefore be send("evm_revert",[1], function(err, result) { ... }); instead of send("evm_revert", function(err, result) { ... });.

Trace with current test implementation

1. timeAdjustement : -50652919
2. evm_increaseTime (line 46)
3. timeAdjustement : -50634919
4. evm_snapshot: (line 75, snapshot 1, timeAdjustement == -50634919)
5. evm_increaseTime (line 77)
6. timeAdjustement : -50616919
7. send("evm_revert", function(err, result) {...}) (line 87)
8. timeAdjustement : -50616919

The current test assert that 4. should be equal to 8, which is false. By specifying snapshot to 1, the timeAdjustment is correctly reverted to the value stored at 4.

[benjamincburns]

There is a mistake in the new test case

But... but I wrote that test case! 😊

Thanks so much for pointing this out, @PhABC - I'll have a look!

[benjamincburns]

I merged this PR, but forgot the --no-ff flag :-(

[graemecode]

Unused var, fyi