Allow interactions to be internal #195
Conversation
Codecov Report
@@ Coverage Diff @@
## add-amounts-to-interaction #195 +/- ##
==============================================================
+ Coverage 64.44% 64.81% +0.37%
==============================================================
Files 189 190 +1
Lines 38988 39147 +159
==============================================================
+ Hits 25125 25375 +250
+ Misses 13863 13772 -91 |
| /// The interacetion should **not** be included in the settlement as | ||
| /// internal buffers will be used instead. | ||
| #[serde(with = "execution_plan_internal")] | ||
| Internal, |
There was a problem hiding this comment.
Just one thought:
maybe, we might want to simulate the transactions with the "later to be internalized AMM interactions", in order to see that the internalization was legit at some point.
If we want to do that, I think its better to also ask the solvers for the executionplancoordinates even for internalized AMMs.
But given that the game of the rules are not yet finalized, I dont' know how important a possible simulation really is.
nlordell
force-pushed
the
add-amounts-to-interaction
branch
from
a4f98eb
to
f4ac00d
Compare
nlordell
force-pushed
the
add-interaction-private-flag
branch
from
5554a2d
to
72145cb
Compare
nlordell
force-pushed
the
add-interaction-private-flag
branch
from
72145cb
to
b1ae8eb
Compare
| .flat_map(|u| &u.execution) | ||
| .all(|u| u.exec_plan.is_some()) | ||
| .flat_map(|u| u.execution.iter().map(|e| &e.exec_plan)); | ||
| let interaction_executions = self.interaction_data.iter().map(|i| &i.exec_plan); |
There was a problem hiding this comment.
This turns out to be a breaking change. The otex solver was not providing an execution plan for its single interaction (it's marked optional on the struct) and therefore stopped being considered in shadow/barn mode and will basically break in prod if we release today.
There was a problem hiding this comment.
Thanks for noticing. Fixed in #237
Now that we have external solvers, we will start requiring them to provide execution/calldata for AMM interactions that they internalize.
This PR changes the
exec_planthat is returned from HTTP solvers to instead use an enum to indicate whether or not it is "internal" or to be included in the execution plan. Currently, all we do with "internal" interactions is to skip them when adding them to the solution and logging a message indicating that it was internalized.Once we start recording settlement competitions in the database, we need to make sure that these internalized AMM interactions get stored somewhere so that they can be verified at a later date. For this, some additional refactoring is necessary in order to make encoding interactions into settlements easier.
One additional point of interest is that we currently don't verify that internalized interactions are indeed over the list of "allowed buffer trading" tokens. I plan on adding this as an follow up PR to not over-complicate this one.
Test Plan
Added a unit test that verifies that internal interactions don't get encoded.