This earlier question asked for the differences between 4 different Git diff strategies, but the only difference that was explained was the difference between myers and patience, which is pretty well explained elsewhere.
How does the histogram strategy work? What differentiates it from patience? The git-diff man page only says that it "extends the patience algorithm to "support low-occurrence common elements"." Other pages mention that it's faster, and that it comes from JGit, but they don't explain where or how its algorithm or results will differ from patience.
Where can I find a description of the histogram algorithm relative to the patience algorithm, with the same level of detail as Bram Cohen's original description of the patience algorithm?
(If it's just a matter of implementation performance with no case that will produce different results, why wasn't it just implemented as a new backend for patience?)