Slow fugitive#filereadable

[SolaWing]

I work on a large repo, and Found that show commit is very slow. I profile vim and found the time cost on git ls-tree, which take several seconds on each call, and is called by fugitive#filereadable, and which is called from other plugin like vim-projectionist, vim-bundler, vim-rails, . following is a profile on show commit from status window:

fugitive-profile.log

As filereadable is a common action to call from anywhere, this problem makes fugitive respond slowly, seriously affecting the user experience.

So my question is: can this performance be optimized to avoid getting the full tree? Or I have to find a way to adjust the caller and not call readable in the fugitive-related buffer?

[tpope]

Short answer is you can opt out with:

let g:fugitive_file_api = v:false

This needs to be documented, but doing that means also documenting how this API works. 🥴

So my question is: can this performance be optimized to avoid getting the full tree?

See #1866 and #1868.