Greg Hurrell

Yes, that's right @Aster89. The way Command-T works (unless configured otherwise) is to recursively try all the possible ways of scoring a given match, and to use the highest-scoring one...

Right, [the Ruby code which is highlighting the string is here](https://github.com/wincent/command-t/blob/6a00a2cabfb5ee3f49ff556ceea56892825c0397/ruby/command-t/lib/command-t/match_window.rb#L402-L424). The memoization is happening in C-land ([see `memo` in this file](https://github.com/wincent/command-t/blob/6a00a2cabfb5ee3f49ff556ceea56892825c0397/ruby/command-t/ext/command-t/match.c#L23), which is what I am calling an "implementation...

Do you have `'autochdir'` set, by any chance? (See `:h 'autochdir'`). That setting causes Vim to change directory every time you open a file, which causes Command-T to have to...

Must be something else going on then. This is with the standard `:CommandT` finder, right? Can you confirm that `:pwd` is stable and not changing? There might be something unknown...

It's not a file; it's an in-memory cache — as in, if you quit Vim, the cache is gone.

You could try disabling all other plugins and starting with an empty `.vimrc`; like I said above, there's nothing internal to Command-T that would cause it to rescan a cached...

> Probably ruby 2.4 isn't compatible. In principle, there's no reason why v2.4 shouldn't work. I'm using different versions of Ruby on different machines (v2.6.3-p62 on the machine I'm on...

Thanks. Looks reasonable!

Yeah, I think so @bsiddiqui. Sorry to be slow to deal with it (busy at work). Will get to it shortly.

Can't really review this on my phone but looks pretty epic. The trie makes sense: trade a bit of overhead for its construction in exchange for cheaper search space pruning...