1 [[generate_patch_text_with_p]]
2 Generating patch text with -p
3 -----------------------------
9 linkgit:git-diff-index[1],
10 linkgit:git-diff-tree[1], or
11 linkgit:git-diff-files[1]
12 with the `-p` option produces patch text.
13 You can customize the creation of patch text via the
14 `GIT_EXTERNAL_DIFF` and the `GIT_DIFF_OPTS` environment variables
15 (see linkgit:git[1]), and the `diff` attribute (see linkgit:gitattributes[5]).
17 What the -p option produces is slightly different from the traditional
20 1. It is preceded by a "git diff" header that looks like this:
22 diff --git a/file1 b/file2
24 The `a/` and `b/` filenames are the same unless rename/copy is
25 involved. Especially, even for a creation or a deletion,
26 `/dev/null` is _not_ used in place of the `a/` or `b/` filenames.
28 When a rename/copy is involved, `file1` and `file2` show the
29 name of the source file of the rename/copy and the name of
30 the file that the rename/copy produces, respectively.
32 2. It is followed by one or more extended header lines:
36 deleted file mode <mode>
42 similarity index <number>
43 dissimilarity index <number>
44 index <hash>..<hash> <mode>
46 File modes are printed as 6-digit octal numbers including the file type
47 and file permission bits.
49 Path names in extended headers do not include the `a/` and `b/` prefixes.
51 The similarity index is the percentage of unchanged lines, and
52 the dissimilarity index is the percentage of changed lines. It
53 is a rounded down integer, followed by a percent sign. The
54 similarity index value of 100% is thus reserved for two equal
55 files, while 100% dissimilarity means that no line from the old
56 file made it into the new one.
58 The index line includes the blob object names before and after the change.
59 The <mode> is included if the file mode does not change; otherwise,
60 separate lines indicate the old and the new mode.
62 3. Pathnames with "unusual" characters are quoted as explained for
63 the configuration variable `core.quotePath` (see
64 linkgit:git-config[1]).
66 4. All the `file1` files in the output refer to files before the
67 commit, and all the `file2` files refer to files after the commit.
68 It is incorrect to apply each change to each file sequentially. For
69 example, this patch will swap a and b:
78 5. Hunk headers mention the name of the function to which the hunk
79 applies. See "Defining a custom hunk-header" in
80 linkgit:gitattributes[5] for details of how to tailor this to
87 Any diff-generating command can take the `-c` or `--cc` option to
88 produce a 'combined diff' when showing a merge. This is the default
89 format when showing merges with linkgit:git-diff[1] or
90 linkgit:git-show[1]. Note also that you can give suitable
91 `--diff-merges` option to any of these commands to force generation of
92 diffs in a specific format.
94 A "combined diff" format looks like this:
97 diff --combined describe.c
98 index fabadb8,cc95eb0..4866510
101 @@@ -98,20 -98,12 +98,20 @@@
102 return (a_date > b_date) ? -1 : (a_date == b_date) ? 0 : 1;
105 - static void describe(char *arg)
106 -static void describe(struct commit *cmit, int last_one)
107 ++static void describe(char *arg, int last_one)
109 + unsigned char sha1[20];
110 + struct commit *cmit;
111 struct commit_list *list;
112 static int initialized = 0;
113 struct commit_name *n;
115 + if (get_sha1(arg, sha1) < 0)
116 + usage(describe_usage);
117 + cmit = lookup_commit_reference(sha1);
119 + usage(describe_usage);
123 for_each_ref(get_name);
126 1. It is preceded by a "git diff" header, that looks like
127 this (when the `-c` option is used):
131 or like this (when the `--cc` option is used):
135 2. It is followed by one or more extended header lines
136 (this example shows a merge with two parents):
138 index <hash>,<hash>..<hash>
139 mode <mode>,<mode>..<mode>
141 deleted file mode <mode>,<mode>
143 The `mode <mode>,<mode>..<mode>` line appears only if at least one of
144 the <mode> is different from the rest. Extended headers with
145 information about detected content movement (renames and
146 copying detection) are designed to work with the diff of two
147 <tree-ish> and are not used by combined diff format.
149 3. It is followed by a two-line from-file/to-file header:
154 Similar to the two-line header for the traditional 'unified' diff
155 format, `/dev/null` is used to signal created or deleted
158 However, if the --combined-all-paths option is provided, instead of a
159 two-line from-file/to-file, you get an N+1 line from-file/to-file header,
160 where N is the number of parents in the merge commit:
167 This extended format can be useful if rename or copy detection is
168 active, to allow you to see the original name of the file in different
171 4. Chunk header format is modified to prevent people from
172 accidentally feeding it to `patch -p1`. Combined diff format
173 was created for review of merge commit changes, and was not
174 meant to be applied. The change is similar to the change in the
175 extended 'index' header:
177 @@@ <from-file-range> <from-file-range> <to-file-range> @@@
179 There are (number of parents + 1) `@` characters in the chunk
180 header for combined diff format.
182 Unlike the traditional 'unified' diff format, which shows two
183 files A and B with a single column that has `-` (minus --
184 appears in A but removed in B), `+` (plus -- missing in A but
185 added to B), or `" "` (space -- unchanged) prefix, this format
186 compares two or more files file1, file2,... with one file X, and
187 shows how X differs from each of fileN. One column for each of
188 fileN is prepended to the output line to note how X's line is
191 A `-` character in the column N means that the line appears in
192 fileN but it does not appear in the result. A `+` character
193 in the column N means that the line appears in the result,
194 and fileN does not have that line (in other words, the line was
195 added, from the point of view of that parent).
197 In the above example output, the function signature was changed
198 from both files (hence two `-` removals from both file1 and
199 file2, plus `++` to mean one line that was added does not appear
200 in either file1 or file2). Also, eight other lines are the same
201 from file1 but do not appear in file2 (hence prefixed with `+`).
203 When shown by `git diff-tree -c`, it compares the parents of a
204 merge commit with the merge result (i.e. file1..fileN are the
205 parents). When shown by `git diff-files -c`, it compares the
206 two unresolved merge parents with the working tree file
207 (i.e. file1 is stage 2 aka "our version", file2 is stage 3 aka