What should we expect about the Tab Autocompletion?

Seriously, since the new Tab Fusion Model has been release, what should we expect from the next improvements?

I am a full stack engineer, and in my daily work, the tab autocompletion has saved a lot of time for me compared with the compose or chat functionality. So I heavily rely on the tab autocompletion.

What should I expect for the improvements of the tab autocompletion?

1、It seems that the tab autocompletion also has its own context(not sure the exact size), but for the latency, the context window size may be not so big. If we open some files, the tab autocompletion seems work better, from that I can conclude the tab autocompletion may consider some context from the opening files.

If we remain only one file open and the file contains few codes, the completion seems cannot generate high quality codes(we may define the high quality codes as good codes that conform to the whole projects). I guess the context for the tab autocompletion model comes from both the current editing file with higher priority and the opening files with lower priority. Since if there are no any other opening files and the current editing file has few codes, it may not perform well.

So, is there any way or possibilities to pre-provide some codes for the tab completion model so that all codes generated will conform to the provided example codes? We may call this codes pre-heat.

2、Lower latency. I know it’s very difficult to keep a very low latency for the tab completion model. To an AI code editor, it seems we don’t have to install many language specific extensions. Compared to these local extensions providing language supporting, the latency of the tab autocompletion is very critical for the coding experience. The latency should not let the coders to wait for some time to show the completion codes. The SuperMaven does well in this point.

Any information about the merge work from SuperMaven?

Anyway, lower latency and larger or controllable context may be the answer for me.

What about you?