Since guilt is heavily based on Mercurial queues (mq), some of the following is shamelessly stolen from the mq page [[1]].

Introduction

Andrew Morton originally developed a set of scripts for maintaining kernel patches outside of any SCM tool. Others extended these into a suite called quilt [[2]]. The basic idea behind quilt is to maintain patches instead of maintaining source files. Patches can be added, removed or reordered, and they can be refreshed as you fix bugs or update to a new base revision. quilt is very powerful, but it is not integrated with the underlying SCM tools. This makes it difficult to visualize your changes.

Guilt allows one to use quilt functionality on top of a Git repository. Changes are maintained as patches which are committed into Git. Commits can be removed or reordered, and the underlying patch can be refreshed based on changes made in the working directory. The patch directory can also be placed under revision control, so you can have a separate history of changes made to your patches.

Patches Directory

In Guilt, all the patches are stored in .git/patches/$branch/, where $branch is the name of the branch being worked on. This means that one can have a independent series of patches for each branch present in the repository. Each of these per-branch directories contains 2 special files:

series: This file contains a list of all the patch filenames relative to the per-branch patch directory. Empty and commented out lines are ignored.

status: This file contains the state of the stack. What patches are applied.

Diving in

This section is mostly for those savvy enough to just dive in, and not read documentation ;) A more detailed description of what is going on is below.

    $ cd some_git_repo

initialize the patch directory

    $ guilt init

create a new patch called 'patch1'

    $ guilt new patch1

edit somefile

    $ vim somefile

refresh patch1 with the changes we just made

    $ guilt refresh

If we look at .git/patches/$branch/patch1, we see a diff with the changes we just made.

Create a new patch called 'patch2'

    $ guilt new patch2

edit somefile again

    $ vim somefile

refresh patch2

    $ guilt refresh

list all applied patches

    $ guilt applied

list all applied and unapplied patches

    $ guilt series

pop the top most applied patch

    $ guilt pop

push the next patch on top of the currently applied patches

    $ guilt push

pop all patches

    $ guilt pop -a

push all patches

    $ guilt push -a

Why use Guilt?

The quilt pdf [[3]] paper seems to be a good starting point for why use a quilt-like system. As to why use Guilt over other quilt-like git porceilans, that's mostly up to:

1. User's taste
2. Format of the patches directory

Working with Guilt:

First, get a repository to work on. Here's one that we'll use as an example:

$ git-clone git://git.kernel.org/pub/scm/linux/kernel/jsipek/guilt-hello.git

Now, it is time to initialize the patches directory using guilt's init command:

$ cd hello
$ guilt init

Create a new patch called "new_hello_string"

$ guilt new new_hello_string

This creates a new empty file .git/patches/master/new_hello_string.

Now, it is time to change hello.c. Suppose we change the string being outputed to "Howdy!\n".

Currently, the changes live in the working copy only. Running refresh refreshes the (currently empty) patch:

$ guilt refresh

If we look at the patch file again, we'll see that it now contains a patch that changes the hello string.

Let's create a second patch called "say_name"

$ guilt new say_name

...and edit hello.c to output the programs name by adding:

printf("My name is '%s'\n", argv[0]);

And of course, we refresh the patch:

$ guilt refresh

As you might have expected, this updates the second patch file with the appropriate diff.