You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Thoughts that might be important for this #48 (comment)
In current implementation we are adding aggregate/add right on the exit of the first queue that is wired with aggregate/queue. This gives us the control of the flow, as well as to guarantee that no further same piece can be added.
However, by executing aggregate/add at that point, receipt chain will not enable storefront to get to know aggregate + inclusion proof. Probably worth to perform the aggregate/add when we include things into an aggregate that is ready to be sent. With that in mind, we would wire the aggregate-add queue when we queue the deal (in dealer-queue workflow). The disadvantage is not being able to re-act fast for multiple piece adds, which given we control the storefront now is not really a problem. However, if we would have direct users of the aggregation pipeline (like what we talked with FVM team), opening this door could be an issue.
An alternative here would be to introduce a third capability aggregate/inclusion, where aggregate/add is executed as in this PR, and the above solution would be the aggregate/inclusion (where of course we would add effect from aggregate/add to aggregate/inclusion.
The text was updated successfully, but these errors were encountered:
Thoughts that might be important for this #48 (comment)
In current implementation we are adding aggregate/add right on the exit of the first queue that is wired with aggregate/queue. This gives us the control of the flow, as well as to guarantee that no further same piece can be added.
However, by executing aggregate/add at that point, receipt chain will not enable storefront to get to know aggregate + inclusion proof. Probably worth to perform the aggregate/add when we include things into an aggregate that is ready to be sent. With that in mind, we would wire the aggregate-add queue when we queue the deal (in dealer-queue workflow). The disadvantage is not being able to re-act fast for multiple piece adds, which given we control the storefront now is not really a problem. However, if we would have direct users of the aggregation pipeline (like what we talked with FVM team), opening this door could be an issue.
An alternative here would be to introduce a third capability aggregate/inclusion, where aggregate/add is executed as in this PR, and the above solution would be the aggregate/inclusion (where of course we would add effect from aggregate/add to aggregate/inclusion.
The text was updated successfully, but these errors were encountered: