Daniel T

Out of interest, what version of Mac? I managed to install ok with homebrew on catalina

> Did not find CMake 'cmake' Found CMake: NO Please try again after `$ brew install cmake`

Hey @jim-collier , sorry for the long delay. There was a bug in the xattr implementation due to mtime rounding, fix at #494. The xattr method is potentially flawed if...

Further on this, have developed a new [optimisation](https://github.com/SeeSpotRun/rmlint/tree/feature/keep_reflinked) to make duplicate detection more efficient if there are existing reflinks. Due to the overhead of `fiemap` parsing, initial benchmarking suggests it's...

> wouldn't the former be more accessible to script-based wrappers though? Definitely, although json is pretty easily parsed too. Another reason for keeping `--xattr` is that [limited] benchmarking indicates that...

Results for cloning 2 x 1 gigabyte dupes on SSD: ``` $ /usr/bin/time ./rmlint --dedupe 1g1 1g2 # initial deduplication 0.00user 3.62system 0:43.50elapsed 8%CPU (0avgtext+0avgdata 4604maxresident)k $ /usr/bin/time ./rmlint --dedupe...

> see if there's a way to make it more robust This approach records both the path and the mtime of the original, as well as the mtime of the...

> does that mean rmlint does actually know if it has already deduped a file before? And if so, any thoughts on why the % already deduped seems much lower...

Ok I think it's time to close this out. A quick summary: 1. The original question was *"Is rmlint with --xattr-** *flags incremental for Btrfs deduplication, or just generating hashes?"...

Should be possible. I don't have much time at the moment but will try to get back to this in a few weeks.