I find that I often have to suffer typing some similar latex codes many times in different equations or the same equation. So it will be convenient that all the codes in the existed equations in the buffer could be used as the backend for autocompletion.
It seems to me that it should be the same technique used in the autocompletion for the plain text. That is,
- collect all the codes/words separated by space in the buffer as the backend
- list the candidates with the same beginning
It's great if some superuser can implement this. Any comments or suggestions are also welcome.