Open JeffreyBenjaminBrown opened 8 years ago
If I understand the issue, it definitely walks the line between a bug and a feature. I actually find this behavior pretty useful in one specific context: I am in a height-1 view and absent-mindedly add a new child to one of the atoms in the list, which is not expanded. I push, and my new child goes away. I tend to make this mistake only when none of the visible atoms has children, so rather than undo-ing in the buffer, copying the child, increasing the height, and then pasting the child in, I just increase the height, then undo and push. It saves some time, but it's OK if this goes away at some point, perhaps by avoiding conflicts by attaching a version number or "last changed" timestamp to the buffer.
Let me know if that's not what you meant. Good luck on your interviews.
This is not a bug, but it is a danger. It arises because (I assume) the undo stack for a buffer starts counting changes before the buffer has initially loaded its contents.