>> block cloning, which allows shallow copies of blocks in file copies. This is optional, and disabled by default; [...] > This sounds interesting, but I wonde

>> block cloning, which allows shallow copies of blocks in file copies. This is optional, and disabled by default; [...] > This sounds interesting, but I wonder what would be the advantages compared to enabling dedup on a specific dataset (disclaimer: I've never used dedup). In a steady-state system the results should be (almost) identical. But deduplication is either an inline process (which has limited data and thus is less efficient) or post-process (which runs maybe once a day and thus takes a while), while a block clone of a file is instantaneous, it takes no space and you don't have to wait for the next deduplication run to get your free space back. Which is helpful for large files (think qemu disk images). If you have ever used a (commercial) file system that implements this, you will never want to go back :)