On Mon, Apr 27, 2020 at 9:10 PM Jeff King <peff@xxxxxxxx> wrote: > Without engaging my brain to think about what this code is doing or > whether there might be clever solutions, it really sounds like we might > consider using this quadratic code for small cases if it produces better > results, and then switching to the less-accurate greedy implementation > when we need to. I remember having the exact same thought at the time I wrote this, but my limited tests with 10k lines files were fine so I discarded it. Falling-back on the greedy algorithm seems reasonable in that case. Antoine
