mpatch maintainer: Chris Mason <chris.mason@oracle.com> http://oss.oracle.com/~mason/mpatch mpatch applies diffs and is generally similar to patch, but it can help resolve a number of common causes of patch rejects. By default, it applies diffs with the same offset and fuzz rules that patch uses, and mpatch is intended to be as safe as patch for normal usage. I'm very interested in any bug reports where mpatch does something unexpected. When run with -a, it will merge the rejects back into the original file and open a merge program (gvimdiff by default) to help you review the changes. mpatch also supports the git patch extensions. Outside of an SCM, these are mostly useful to preserve permissions changes in a patch. Reject merging works by using the entire hunk as a source of context, and being more relaxed about finding extra lines in the source file, or not finding every line from the hunk. If the hunk tries to delete a line that is not present in the file, mpatch will put a deletion marker into the merged result so you can fix things up. Reject files are always created in the patch style, allowing you to verify the result of the merge. Rejects created by mpatch are unified diffs, and have a simple patch header. vimdiff users may want to look at the macros in vimrc. These make it easy to copy conflict sections from one vimdiff window to another. quilt users will want to look at cmd/qp (this is not installed by default). It is a helper script to parse the output from quilt push looking for fuzz and rejects, and it provides a prompt that will run mpatch for you. Mercurial users will want to look at cmd/mp (this is not installed by default). It is a helper script to parse the output from hg qpush looking for fuzz and rejects, and it provides a prompt that will run mpatch for you. Sample usage: Apply diff and auto merge any rejects found mpatch -a < diff Auto merge the reject hunks from file.rej into file (file.rej is automatically tried if not provided on the command line) mpatch -a file [file.rej] Auto merge diff into file (useful when diff doesn't have headers) mpatch -a file diff Auto merge diff into file, but store the results in 'otherfile' mpatch -o otherfile file diff Other merge programs are allowed: mpatch -a -m kdiff3 < diff You can use any program that can accept two file names on the command line. Other options are documented in mpatch -h