rej tries to merge simple patch-rejects and then run a merge program so the
changes can be verified.  It is not meant to resolve complex problems that
would not be immediately obvious to the programmer, the goal instead is to
quickly fix the easy problems.

rej understands both unified and context diffs.

There are four basic rejects fixable via rej.

  1. missing context at the top or bottom of the hunk
  2. different context in the middle of the hunk
  3. slightly different lines removed by the hunk than exist in the file
  4. Large hunks that might apply if they were broken up into smaller ones

rej also allows you to tag hunks in a reject with special processing
hints.