Out of simple curiosity, how does 'git annex migrate' work? I'm mostly wondering how file integrity is ensured.
Let's say you want to migrate foo.txt from (say) md5 to sha256. Does git annex simply sha256sum foo.txt and rename it in the .git/objects folder to the new sum? Or does it md5sum foo.txt, verify that it's the not corrupt, then sha256sum it and rename to the new sum?
You could run into problems if you migrate without first verifying; if the file is corrupt and you simply sha256sum and rename it, then the file wouldn't seem corrupt at your next fsck.
I'm sure you've considered this during the basic design phase of git-annex, but I'd just like to be sure. I'm kind of paranoid when it comes to data integrity. =P