feat: Ekubo Integration

[die-herdplatte]

In conjunction with propeller-heads/tycho-protocol-sdk#172 and propeller-heads/tycho-simulation#173

[die-herdplatte]

Should Ekubo's callbacks be directly integrated into TychoRouter.sol?

[tamaralipows]

Thanks a lot for this PR!

Regarding this:

Should Ekubo's callbacks be directly integrated into TychoRouter.sol?

This is discouraged if you can make the callback work with the handleCallback method in the executor - which seems to be the case here. No changes to the main TychoRouter necessary.

Could you please update your branch with the main branch - we have recently pushed a fix that should make CI run properly in external contribution PRs now.

Other than this - I would recommend to add an integration test on for the executor contract which uses data encoded by the swap encoder. Would it be possible for you to add one for the EkuboExecutor? See simple examples here:

testSwapIntegration in BalancerV2ExecutorTest
testMultipleSwapIntegration in UniswapV4ExecutorTest

[die-herdplatte]

This is discouraged if you can make the callback work with the handleCallback method in the executor - which seems to be the case here. No changes to the main TychoRouter necessary.

Slightly related: Do I need the direct callback implementations on the executor itself (locked and payCallback).
I just implemented them because other executors did the same and it simplifies tests but shouldn't the executor only be callable by the router?

Could you please update your branch with the main branch - we have recently pushed a fix that should make CI run properly in external contribution PRs now.

Done, but seems like there are still issues.

[tamaralipows]

You're right to point out that the executor will only be callable through delegatecalls by the router, and thus those direct callbacks don't need to be implemented. We had included them just for clarity for the user and ease of testing - it's fine if you leave them here though.

One thing that I'd like to double check - does this work through the router? We need to make sure the callback data starts with the executor address in that case:

    function _handleCallback(bytes calldata data) internal {
        address executor = address(uint160(bytes20(data[data.length - 20:])));

I'll take over this PR to double check this - thanks!

About the CI - this should be fixed soon, we are working on this internally.

[die-herdplatte]

I just noticed that the executor doesn't use permits to pay Ekubo Core.
I can rectify this until tomorrow if you want.

[tamaralipows]

I just noticed that the executor doesn't use permits to pay Ekubo Core.
I can rectify this until tomorrow if you want.

Sure - that would be a big help!

[die-herdplatte]

Nevermind, I see that the router uses the permit to transfer tokens to itself before calling the executor's swap function.
Then everything is fine from my side now.