Hacker News Reader

That's close to a gig of disk reads; I trust you didn't try ugrep first and then cindex second, without taking into account caching.

I ran both multiple times, alternating (and making sure to clean out the indexes in between). Results were reasonably consistent across runs.

If you're gonna go the csearch route, you should also consider hound. I use it many times per day.

It creates per-directory index files on its first run. ugrep-indexer is also labeled as beta. A couple of relevant quotes from its GitHub site:

“Indexing adds a hidden index file ._UG#_Store to each directory indexed.”

“Re-indexing is incremental, so it will not take as much time as the initial indexing process.”