How to structure a new programming project to benefit the most from llama.vim FIM ?

[MartyLake]

Hello,

Please forgive my ignorance and stop me where I am wrong.

As far as I understand how llama.vim FIM works, it sends chunks of visited files on autocmds: BufEnter/BufLeave/BufWritePost. This makes it possible to have FIM which has in its context the current file, and previously visited files.

organize the project to have one giant file

So one possibility to get the most from llama.vim FIM, would be to have one giant file with the whole project or feature, then I would be sure that llama.vim is able to get the most context when FIM, isn’t it?

open all included files when opening the file of interest

Another possibility, for languages that have file-level include directives (c/cpp/php/etc.), it to open the file of interest, browse to included files to trigger BufEnter, then go back to my main file while llama.vim continues to send more context, isn’t it? If that works, I wonder if it would be out-of-scope to follow include directives by relying on vim’s path system, delegating to the user the responsibility to have a correct include path setup. Or maybe it could be some extension/function that, when called, do the BufEnter dance for each included files ?

other?

Maybe there are other ways to organize a programming project. For example, having a file that contains the general desired architecture, and open this one too so that FIM has some hint of where to go in a broader context ? if so, in which format should that information be written into ?