What diff algorithm relies on finding the longest common prefix and suffix and cursing on the middle substring?

Question

I can't remember where I read this years ago, but there's this variant of a diff algorithm for 2 lines, that finds the longest common prefix and suffix for the 2 lines, and then recurses on the remaining portion.

There might have been some kind of pivot element involved as well. I can't seem to find it on Google when I searched using the above description.

Is there a name for this diff algorithm?