1 Generating patch text with -p
2 -----------------------------
8 linkgit:git-diff-index[1],
9 linkgit:git-diff-tree[1], or
10 linkgit:git-diff-files[1]
11 with the `-p` option produces patch text.
12 You can customize the creation of patch text via the
13 `GIT_EXTERNAL_DIFF` and the `GIT_DIFF_OPTS` environment variables.
15 What the -p option produces is slightly different from the traditional
18 1. It is preceded with a "git diff" header that looks like this:
20 diff --git a/file1 b/file2
22 The `a/` and `b/` filenames are the same unless rename/copy is
23 involved. Especially, even for a creation or a deletion,
24 `/dev/null` is _not_ used in place of the `a/` or `b/` filenames.
26 When rename/copy is involved, `file1` and `file2` show the
27 name of the source file of the rename/copy and the name of
28 the file that rename/copy produces, respectively.
30 2. It is followed by one or more extended header lines:
34 deleted file mode <mode>
40 similarity index <number>
41 dissimilarity index <number>
42 index <hash>..<hash> <mode>
44 File modes are printed as 6-digit octal numbers including the file type
45 and file permission bits.
47 Path names in extended headers do not include the `a/` and `b/` prefixes.
49 The similarity index is the percentage of unchanged lines, and
50 the dissimilarity index is the percentage of changed lines. It
51 is a rounded down integer, followed by a percent sign. The
52 similarity index value of 100% is thus reserved for two equal
53 files, while 100% dissimilarity means that no line from the old
54 file made it into the new one.
56 The index line includes the blob object names before and after the change.
57 The <mode> is included if the file mode does not change; otherwise,
58 separate lines indicate the old and the new mode.
60 3. Pathnames with "unusual" characters are quoted as explained for
61 the configuration variable `core.quotePath` (see
62 linkgit:git-config[1]).
64 4. All the `file1` files in the output refer to files before the
65 commit, and all the `file2` files refer to files after the commit.
66 It is incorrect to apply each change to each file sequentially. For
67 example, this patch will swap a and b:
80 Any diff-generating command can take the `-c` or `--cc` option to
81 produce a 'combined diff' when showing a merge. This is the default
82 format when showing merges with linkgit:git-diff[1] or
83 linkgit:git-show[1]. Note also that you can give the `-m` option to any
84 of these commands to force generation of diffs with individual parents
87 A "combined diff" format looks like this:
90 diff --combined describe.c
91 index fabadb8,cc95eb0..4866510
94 @@@ -98,20 -98,12 +98,20 @@@
95 return (a_date > b_date) ? -1 : (a_date == b_date) ? 0 : 1;
98 - static void describe(char *arg)
99 -static void describe(struct commit *cmit, int last_one)
100 ++static void describe(char *arg, int last_one)
102 + unsigned char sha1[20];
103 + struct commit *cmit;
104 struct commit_list *list;
105 static int initialized = 0;
106 struct commit_name *n;
108 + if (get_sha1(arg, sha1) < 0)
109 + usage(describe_usage);
110 + cmit = lookup_commit_reference(sha1);
112 + usage(describe_usage);
116 for_each_ref(get_name);
119 1. It is preceded with a "git diff" header, that looks like
120 this (when the `-c` option is used):
124 or like this (when the `--cc` option is used):
128 2. It is followed by one or more extended header lines
129 (this example shows a merge with two parents):
131 index <hash>,<hash>..<hash>
132 mode <mode>,<mode>..<mode>
134 deleted file mode <mode>,<mode>
136 The `mode <mode>,<mode>..<mode>` line appears only if at least one of
137 the <mode> is different from the rest. Extended headers with
138 information about detected contents movement (renames and
139 copying detection) are designed to work with diff of two
140 <tree-ish> and are not used by combined diff format.
142 3. It is followed by two-line from-file/to-file header
147 Similar to two-line header for traditional 'unified' diff
148 format, `/dev/null` is used to signal created or deleted
151 However, if the --combined-all-paths option is provided, instead of a
152 two-line from-file/to-file you get a N+1 line from-file/to-file header,
153 where N is the number of parents in the merge commit
160 This extended format can be useful if rename or copy detection is
161 active, to allow you to see the original name of the file in different
164 4. Chunk header format is modified to prevent people from
165 accidentally feeding it to `patch -p1`. Combined diff format
166 was created for review of merge commit changes, and was not
167 meant to be applied. The change is similar to the change in the
168 extended 'index' header:
170 @@@ <from-file-range> <from-file-range> <to-file-range> @@@
172 There are (number of parents + 1) `@` characters in the chunk
173 header for combined diff format.
175 Unlike the traditional 'unified' diff format, which shows two
176 files A and B with a single column that has `-` (minus --
177 appears in A but removed in B), `+` (plus -- missing in A but
178 added to B), or `" "` (space -- unchanged) prefix, this format
179 compares two or more files file1, file2,... with one file X, and
180 shows how X differs from each of fileN. One column for each of
181 fileN is prepended to the output line to note how X's line is
184 A `-` character in the column N means that the line appears in
185 fileN but it does not appear in the result. A `+` character
186 in the column N means that the line appears in the result,
187 and fileN does not have that line (in other words, the line was
188 added, from the point of view of that parent).
190 In the above example output, the function signature was changed
191 from both files (hence two `-` removals from both file1 and
192 file2, plus `++` to mean one line that was added does not appear
193 in either file1 or file2). Also eight other lines are the same
194 from file1 but do not appear in file2 (hence prefixed with `+`).
196 When shown by `git diff-tree -c`, it compares the parents of a
197 merge commit with the merge result (i.e. file1..fileN are the
198 parents). When shown by `git diff-files -c`, it compares the
199 two unresolved merge parents with the working tree file
200 (i.e. file1 is stage 2 aka "our version", file2 is stage 3 aka