nothing should happen in overlay db regarding keyspace, do resolution
in block_execution from client.
Consider removing KeySpace from backend InMemory to (kv will get it).

will be good.

values is a wrong direction (it would mean historying things at two
levels), so keeping it in kv seems better.
Still in transaction we want this keyspace from kv (kv transaction is
not exploitable).

vec for compactness in state_db (previously we where removing duplicate
with inner map, now we keep map up to).

value).

(here we never revert canonical).

block modification and worst could erase changes).

kaichao authored Oct 05, 2019 and Shawn Tabrizi committed Oct 04, 2019

* Attemp to fix it.

* Move env to the begining.

* babe: fix type on find_pre_digest call

* fork-tree: optimize prune

* babe: fix test compilation

* babe: prune epoch tree when importing a new epoch change

* fork-tree: fix tree pruning

* babe: actually prune epoch change fork tree

* Fix typos

* babe: add test for epoch tree pruning

* fork-tree: fix pruning of stale forks

* Implement and test batch

* Add files.

* Remove comments.

* Update srml/utility/src/lib.rs

Co-Authored-By: Bastian Köcher <[email protected]>

* Fixes