On Fri, Apr 1, 2022 at 10:37 AM Stefan Sperling <s...@elego.de> wrote: > > On Fri, Apr 01, 2022 at 04:17:45PM +0200, Johan Corveleyn wrote: > > - Perhaps the fundamental diff algorithm in SVN is fine, but it has a > > performance bug / low-level inefficiency? I think that should be > > explored first, because it might fix most of this problem without > > requiring a discussion about the precise output, and without huge > > efforts of rewrites. > > I myself find the current SVN diff code very difficult to make sense of. > Maybe someone smarter than me (like Sander Striker?) could figure this out.
Just a reminder that there's a program in tools/dev/gen-test-data that can generate files that trigger the slow diff phenomenon. Cheers, Nathan
