akiradeveloper / ijk Goto Github PK

In vi, user is able to repeat a command multiple times by num followed by the command (e.g. 4x to remove the character 4 times).

Currently, recognizing the num part is only to JUMP node on FSM. To make it possible for num part to cooperate with other commands, we may need to move the num part to INIT node, or add new edge from JUMP node (should be renamed to NUM btw).

Anyway, the change should be undoable atomically and the concept of change log set is implemented in prior.

The current ijk does not implement . to repeat the latest command because simply repeating the latest log may be invalid. For example, delete 4 chars when there is only 3 chars in the buffer, what would happen? This may lead to SEGV.

An idea is that apply_log should return Result and when it is Err the editor state is restored and the repeat command only succeed the log returns Ok.

When shifting multiple lines with range+> or <, the undoing will undo the change line by line, no atomically. The natural behavior should be undoing atomically. Obviously we need a concept like change log set which is namely a set of change log which is undoable at once.

The change log set is, in database context, kind of transaction which contains multiple change set which is able to revert atomically later.

As vi user, it is kind a weird to lack gg command but forced to 1G to jump to the top. But today I think it should have.

C-f does not reach the end when there is shorter than one page distance to the end. This behavior is bit awkward compared to original vi. The cursor should reach the end as G jumps to the end. (pressing C-f continuously = G) So is C-b.