Klever

Now we almost finish a new error trace format. There are 2 new entities: actions and callback actions. The first entity is represented as an array of comments where array indexes correspond to action identifiers. Edges belonging to corresponding actions are additionally marked with attribute action. The second entity is represented as a set of identifiers of actions of the first kind.

Actions aren't notes and their visualization does differ since one action can cover multiple edges. Ilja suggested to visualize actions like auxiliary functions with notes. But this isn't good enough. It does has sense to increase offsets for edges of actions by 1 and to have narrows to show/hide these edges like for functions. But edges with notes and warnings that lie below actions in their stacks shouldn't be hidden like for nested notes.

Callback actions shouldn't be distinguished much here except corresponding comments can be bold. They have much sense for another issue I will open soon.