1GIT-DIFF(1) Git Manual GIT-DIFF(1)
2
6 git-diff - Show changes between commits, commit and working tree, etc
7
9 git diff [<common diff options>] <commit>{0,2} [--] [<path>...]
10
12 Show changes between two trees, a tree and the working tree, a tree and
13 the index file, or the index file and the working tree.
14 git diff [--options] [--] [<path>...]
16 This form is to view the changes you made relative to the index
17 (staging area for the next commit). In other words, the differences
18 are what you could tell git to further add to the index but you
19 still haven’t. You can stage these changes by using git-add(1).
20 If exactly two paths are given, and at least one is untracked,
22 compare the two files / directories. This behavior can be forced by
23 --no-index.
24 git diff [--options] --cached [<commit>] [--] [<path>...]
26 This form is to view the changes you staged for the next commit
27 relative to the named <commit>. Typically you would want comparison
28 with the latest commit, so if you do not give <commit>, it defaults
29 to HEAD. --staged is a synonym of --cached.
30 git diff [--options] <commit> [--] [<path>...]
32 This form is to view the changes you have in your working tree
33 relative to the named <commit>. You can use HEAD to compare it with
34 the latest commit, or a branch name to compare with the tip of a
35 different branch.
36 git diff [--options] <commit> <commit> [--] [<path>...]
38 This is to view the changes between two arbitrary <commit>.
39 git diff [--options] <commit>..<commit> [--] [<path>...]
41 This is synonymous to the previous form. If <commit> on one side is
42 omitted, it will have the same effect as using HEAD instead.
43 git diff [--options] <commit>...<commit> [--] [<path>...]
45 This form is to view the changes on the branch containing and up to
46 the second <commit>, starting at a common ancestor of both
47 <commit>. "git diff A...B" is equivalent to "git diff
48 $(git-merge-base A B) B". You can omit any one of <commit>, which
49 has the same effect as using HEAD instead.
50 Just in case if you are doing something exotic, it should be noted that
52 all of the <commit> in the above description, except for the last two
53 forms that use ".." notations, can be any <tree-ish>.
54 For a more complete list of ways to spell <commit>, see "SPECIFYING
56 REVISIONS" section in git-rev-parse(1). However, "diff" is about
57 comparing two endpoints, not ranges, and the range notations
58 ("<commit>..<commit>" and "<commit>...<commit>") do not mean a range as
59 defined in the "SPECIFYING RANGES" section in git-rev-parse(1).
60
62 -p, -u
63 Generate patch (see section on generating patches). This is the
64 default.
65 -U<n>, --unified=<n>
67 Generate diffs with <n> lines of context instead of the usual
68 three. Implies -p.
69 --raw
71 Generate the raw format.
72 --patch-with-raw
74 Synonym for -p --raw.
75 --patience
77 Generate a diff using the "patience diff" algorithm.
78 --stat[=width[,name-width]]
80 Generate a diffstat. You can override the default output width for
81 80-column terminal by --stat=width. The width of the filename part
82 can be controlled by giving another width to it separated by a
83 comma.
84 --numstat
86 Similar to --stat, but shows number of added and deleted lines in
87 decimal notation and pathname without abbreviation, to make it more
88 machine friendly. For binary files, outputs two - instead of saying
89 0 0.
90 --shortstat
92 Output only the last line of the --stat format containing total
93 number of modified files, as well as number of added and deleted
94 lines.
95 --dirstat[=limit]
97 Output the distribution of relative amount of changes (number of
98 lines added or removed) for each sub-directory. Directories with
99 changes below a cut-off percent (3% by default) are not shown. The
100 cut-off percent can be set with --dirstat=limit. Changes in a child
101 directory is not counted for the parent directory, unless
102 --cumulative is used.
103 --dirstat-by-file[=limit]
105 Same as --dirstat, but counts changed files instead of lines.
106 --summary
108 Output a condensed summary of extended header information such as
109 creations, renames and mode changes.
110 --patch-with-stat
112 Synonym for -p --stat.
113 -z
115 When --raw, --numstat, --name-only or --name-status has been given,
116 do not munge pathnames and use NULs as output field terminators.
117 Without this option, each pathname output will have TAB, LF, double
119 quotes, and backslash characters replaced with \t, \n, \", and \\,
120 respectively, and the pathname will be enclosed in double quotes if
121 any of those replacements occurred.
122 --name-only
124 Show only names of changed files.
125 --name-status
127 Show only names and status of changed files. See the description of
128 the --diff-filter option on what the status letters mean.
129 --submodule[=<format>]
131 Chose the output format for submodule differences. <format> can be
132 one of short and log. short just shows pairs of commit names, this
133 format is used when this option is not given. log is the default
134 value for this option and lists the commits in that commit range
135 like the summary option of git-submodule(1) does.
136 --color[=<when>]
138 Show colored diff. The value must be always (the default), never,
139 or auto.
140 --no-color
142 Turn off colored diff, even when the configuration file gives the
143 default to color output. Same as --color=never.
144 --color-words[=<regex>]
146 Show colored word diff, i.e., color words which have changed. By
147 default, words are separated by whitespace.
148 When a <regex> is specified, every non-overlapping match of the
150 <regex> is considered a word. Anything between these matches is
151 considered whitespace and ignored(!) for the purposes of finding
152 differences. You may want to append |[^[:space:]] to your regular
153 expression to make sure that it matches all non-whitespace
154 characters. A match that contains a newline is silently
155 truncated(!) at the newline.
156 The regex can also be set via a diff driver or configuration
158 option, see gitattributes(1) or git-config(1). Giving it explicitly
159 overrides any diff driver or configuration setting. Diff drivers
160 override configuration settings.
161 --no-renames
163 Turn off rename detection, even when the configuration file gives
164 the default to do so.
165 --check
167 Warn if changes introduce trailing whitespace or an indent that
168 uses a space before a tab. Exits with non-zero status if problems
169 are found. Not compatible with --exit-code.
170 --full-index
172 Instead of the first handful of characters, show the full pre- and
173 post-image blob object names on the "index" line when generating
174 patch format output.
175 --binary
177 In addition to --full-index, output a binary diff that can be
178 applied with git-apply.
179 --abbrev[=<n>]
181 Instead of showing the full 40-byte hexadecimal object name in
182 diff-raw format output and diff-tree header lines, show only a
183 partial prefix. This is independent of the --full-index option
184 above, which controls the diff-patch output format. Non default
185 number of digits can be specified with --abbrev=<n>.
186 -B
188 Break complete rewrite changes into pairs of delete and create.
189 -M
191 Detect renames.
192 -C
194 Detect copies as well as renames. See also --find-copies-harder.
195 --diff-filter=[ACDMRTUXB*]
197 Select only files that are Added (A), Copied (C), Deleted (D),
198 Modified (M), Renamed (R), have their type (i.e. regular file,
199 symlink, submodule, ...) changed (T), are Unmerged (U), are Unknown
200 (X), or have had their pairing Broken (B). Any combination of the
201 filter characters may be used. When * (All-or-none) is added to the
202 combination, all paths are selected if there is any file that
203 matches other criteria in the comparison; if there is no file that
204 matches other criteria, nothing is selected.
205 --find-copies-harder
207 For performance reasons, by default, -C option finds copies only if
208 the original file of the copy was modified in the same changeset.
209 This flag makes the command inspect unmodified files as candidates
210 for the source of copy. This is a very expensive operation for
211 large projects, so use it with caution. Giving more than one -C
212 option has the same effect.
213 -l<num>
215 The -M and -C options require O(n^2) processing time where n is the
216 number of potential rename/copy targets. This option prevents
217 rename/copy detection from running if the number of rename/copy
218 targets exceeds the specified number.
219 -S<string>
221 Look for differences that introduce or remove an instance of
222 <string>. Note that this is different than the string simply
223 appearing in diff output; see the pickaxe entry in gitdiffcore(7)
224 for more details.
225 --pickaxe-all
227 When -S finds a change, show all the changes in that changeset, not
228 just the files that contain the change in <string>.
229 --pickaxe-regex
231 Make the <string> not a plain string but an extended POSIX regex to
232 match.
233 -O<orderfile>
235 Output the patch in the order specified in the <orderfile>, which
236 has one shell glob pattern per line.
237 -R
239 Swap two inputs; that is, show differences from index or on-disk
240 file to tree contents.
241 --relative[=<path>]
243 When run from a subdirectory of the project, it can be told to
244 exclude changes outside the directory and show pathnames relative
245 to it with this option. When you are not in a subdirectory (e.g. in
246 a bare repository), you can name which subdirectory to make the
247 output relative to by giving a <path> as an argument.
248 -a, --text
250 Treat all files as text.
251 --ignore-space-at-eol
253 Ignore changes in whitespace at EOL.
254 -b, --ignore-space-change
256 Ignore changes in amount of whitespace. This ignores whitespace at
257 line end, and considers all other sequences of one or more
258 whitespace characters to be equivalent.
259 -w, --ignore-all-space
261 Ignore whitespace when comparing lines. This ignores differences
262 even if one line has whitespace where the other line has none.
263 --inter-hunk-context=<lines>
265 Show the context between diff hunks, up to the specified number of
266 lines, thereby fusing hunks that are close to each other.
267 --exit-code
269 Make the program exit with codes similar to diff(1). That is, it
270 exits with 1 if there were differences and 0 means no differences.
271 --quiet
273 Disable all output of the program. Implies --exit-code.
274 --ext-diff
276 Allow an external diff helper to be executed. If you set an
277 external diff driver with gitattributes(5), you need to use this
278 option with git-log(1) and friends.
279 --no-ext-diff
281 Disallow external diff drivers.
282 --ignore-submodules
284 Ignore changes to submodules in the diff generation.
285 --src-prefix=<prefix>
287 Show the given source prefix instead of "a/".
288 --dst-prefix=<prefix>
290 Show the given destination prefix instead of "b/".
291 --no-prefix
293 Do not show any source or destination prefix.
294 For more detailed explanation on these common options, see also
296 gitdiffcore(7).
297 <path>...
299 The <paths> parameters, when given, are used to limit the diff to
300 the named paths (you can give directory names and get diff for all
301 files under them).
302
304 The raw output format from "git-diff-index", "git-diff-tree",
305 "git-diff-files" and "git diff --raw" are very similar.
306 These commands all compare two sets of things; what is compared
308 differs:
309 git-diff-index <tree-ish>
311 compares the <tree-ish> and the files on the filesystem.
312 git-diff-index --cached <tree-ish>
314 compares the <tree-ish> and the index.
315 git-diff-tree [-r] <tree-ish-1> <tree-ish-2> [<pattern>...]
317 compares the trees named by the two arguments.
318 git-diff-files [<pattern>...]
320 compares the index and the files on the filesystem.
321 The "git-diff-tree" command begins its output by printing the hash of
323 what is being compared. After that, all the commands print one output
324 line per changed file.
325 An output line is formatted this way:
327 in-place edit :100644 100644 bcd1234... 0123456... M file0
329 copy-edit :100644 100644 abcd123... 1234567... C68 file1 file2
330 rename-edit :100644 100644 abcd123... 1234567... R86 file1 file3
331 create :000000 100644 0000000... 1234567... A file4
332 delete :100644 000000 1234567... 0000000... D file5
333 unmerged :000000 000000 0000000... 0000000... U file6
334 That is, from the left to the right:
337 1. a colon.
339 2. mode for "src"; 000000 if creation or unmerged.
341 3. a space.
343 4. mode for "dst"; 000000 if deletion or unmerged.
345 5. a space.
347 6. sha1 for "src"; 0{40} if creation or unmerged.
349 7. a space.
351 8. sha1 for "dst"; 0{40} if creation, unmerged or "look at work tree".
353 9. a space.
355 10. status, followed by optional "score" number.
357 11. a tab or a NUL when -z option is used.
359 12. path for "src"
361 13. a tab or a NUL when -z option is used; only exists for C or R.
363 14. path for "dst"; only exists for C or R.
365 15. an LF or a NUL when -z option is used, to terminate the record.
367 Possible status letters are:
369 · A: addition of a file
371 · C: copy of a file into a new one
373 · D: deletion of a file
375 · M: modification of the contents or mode of a file
377 · R: renaming of a file
379 · T: change in the type of the file
381 · U: file is unmerged (you must complete the merge before it can be
383 committed)
384 · X: "unknown" change type (most probably a bug, please report it)
386 Status letters C and R are always followed by a score (denoting the
388 percentage of similarity between the source and target of the move or
389 copy), and are the only ones to be so.
390 <sha1> is shown as all 0’s if a file is new on the filesystem and it is
392 out of sync with the index.
393 Example:
395 :100644 100644 5be4a4...... 000000...... M file.c
397 When -z option is not used, TAB, LF, and backslash characters in
400 pathnames are represented as \t, \n, and \\, respectively.
401
403 "git-diff-tree", "git-diff-files" and "git-diff --raw" can take -c or
404 --cc option to generate diff output also for merge commits. The output
405 differs from the format described above in the following way:
406 1. there is a colon for each parent
408 2. there are more "src" modes and "src" sha1
410 3. status is concatenated status characters for each parent
412 4. no optional "score" number
414 5. single path, only for "dst"
416 Example:
418 ::100644 100644 100644 fabadb8... cc95eb0... 4866510... MM describe.c
420 Note that combined diff lists only files which were modified from all
423 parents.
424
426 When "git-diff-index", "git-diff-tree", or "git-diff-files" are run
427 with a -p option, "git diff" without the --raw option, or "git log"
428 with the "-p" option, they do not produce the output described above;
429 instead they produce a patch file. You can customize the creation of
430 such patches via the GIT_EXTERNAL_DIFF and the GIT_DIFF_OPTS
431 environment variables.
432 What the -p option produces is slightly different from the traditional
434 diff format.
435 1. It is preceded with a "git diff" header, that looks like this:
437 diff --git a/file1 b/file2
439 The a/ and b/ filenames are the same unless rename/copy is
441 involved. Especially, even for a creation or a deletion, /dev/null
442 is not used in place of a/ or b/ filenames.
443 When rename/copy is involved, file1 and file2 show the name of the
445 source file of the rename/copy and the name of the file that
446 rename/copy produces, respectively.
447 2. It is followed by one or more extended header lines:
449 old mode <mode>
451 new mode <mode>
452 deleted file mode <mode>
453 new file mode <mode>
454 copy from <path>
455 copy to <path>
456 rename from <path>
457 rename to <path>
458 similarity index <number>
459 dissimilarity index <number>
460 index <hash>..<hash> <mode>
461 3. TAB, LF, double quote and backslash characters in pathnames are
463 represented as \t, \n, \" and \\, respectively. If there is need
464 for such substitution then the whole pathname is put in double
465 quotes.
466 The similarity index is the percentage of unchanged lines, and the
468 dissimilarity index is the percentage of changed lines. It is a rounded
469 down integer, followed by a percent sign. The similarity index value of
470 100% is thus reserved for two equal files, while 100% dissimilarity
471 means that no line from the old file made it into the new one.
472
474 "git-diff-tree", "git-diff-files" and "git-diff" can take -c or --cc
475 option to produce combined diff. For showing a merge commit with "git
476 log -p", this is the default format; you can force showing full diff
477 with the -m option. A combined diff format looks like this:
478 diff --combined describe.c
480 index fabadb8,cc95eb0..4866510
481 --- a/describe.c
482 +++ b/describe.c
483 @@@ -98,20 -98,12 +98,20 @@@
484 return (a_date > b_date) ? -1 : (a_date == b_date) ? 0 : 1;
485 }
486 - static void describe(char *arg)
488 -static void describe(struct commit *cmit, int last_one)
489 ++static void describe(char *arg, int last_one)
490 {
491 + unsigned char sha1[20];
492 + struct commit *cmit;
493 struct commit_list *list;
494 static int initialized = 0;
495 struct commit_name *n;
496 + if (get_sha1(arg, sha1) < 0)
498 + usage(describe_usage);
499 + cmit = lookup_commit_reference(sha1);
500 + if (!cmit)
501 + usage(describe_usage);
502 +
503 if (!initialized) {
504 initialized = 1;
505 for_each_ref(get_name);
506 1. It is preceded with a "git diff" header, that looks like this (when
510 -c option is used):
511 diff --combined file
513 or like this (when --cc option is used):
515 diff --cc file
517 2. It is followed by one or more extended header lines (this example
519 shows a merge with two parents):
520 index <hash>,<hash>..<hash>
522 mode <mode>,<mode>..<mode>
523 new file mode <mode>
524 deleted file mode <mode>,<mode>
525 The mode <mode>,<mode>..<mode> line appears only if at least one of
527 the <mode> is different from the rest. Extended headers with
528 information about detected contents movement (renames and copying
529 detection) are designed to work with diff of two <tree-ish> and are
530 not used by combined diff format.
531 3. It is followed by two-line from-file/to-file header
533 --- a/file
535 +++ b/file
536 Similar to two-line header for traditional unified diff format,
538 /dev/null is used to signal created or deleted files.
539 4. Chunk header format is modified to prevent people from accidentally
541 feeding it to patch -p1. Combined diff format was created for
542 review of merge commit changes, and was not meant for apply. The
543 change is similar to the change in the extended index header:
544 @@@ <from-file-range> <from-file-range> <to-file-range> @@@
546 There are (number of parents + 1) @ characters in the chunk header
548 for combined diff format.
549 Unlike the traditional unified diff format, which shows two files A and
551 B with a single column that has - (minus — appears in A but removed in
552 B), + (plus — missing in A but added to B), or " " (space — unchanged)
553 prefix, this format compares two or more files file1, file2,... with
554 one file X, and shows how X differs from each of fileN. One column for
555 each of fileN is prepended to the output line to note how X’s line is
556 different from it.
557 A - character in the column N means that the line appears in fileN but
559 it does not appear in the result. A + character in the column N means
560 that the line appears in the result, and fileN does not have that line
561 (in other words, the line was added, from the point of view of that
562 parent).
563 In the above example output, the function signature was changed from
565 both files (hence two - removals from both file1 and file2, plus ++ to
566 mean one line that was added does not appear in either file1 nor
567 file2). Also eight other lines are the same from file1 but do not
568 appear in file2 (hence prefixed with +).
569 When shown by git diff-tree -c, it compares the parents of a merge
571 commit with the merge result (i.e. file1..fileN are the parents). When
572 shown by git diff-files -c, it compares the two unresolved merge
573 parents with the working tree file (i.e. file1 is stage 2 aka "our
574 version", file2 is stage 3 aka "their version").
575
577 The --summary option describes newly added, deleted, renamed and copied
578 files. The --stat option adds diffstat(1) graph to the output. These
579 options can be combined with other options, such as -p, and are meant
580 for human consumption.
581 When showing a change that involves a rename or a copy, --stat output
583 formats the pathnames compactly by combining common prefix and suffix
584 of the pathnames. For example, a change that moves arch/i386/Makefile
585 to arch/x86/Makefile while modifying 4 lines will be shown like this:
586 arch/{i386 => x86}/Makefile | 4 +--
588 The --numstat option gives the diffstat(1) information but is designed
591 for easier machine consumption. An entry in --numstat output looks like
592 this:
593 1 2 README
595 3 1 arch/{i386 => x86}/Makefile
596 That is, from left to right:
599 1. the number of added lines;
601 2. a tab;
603 3. the number of deleted lines;
605 4. a tab;
607 5. pathname (possibly with rename/copy information);
609 6. a newline.
611 When -z output option is in effect, the output is formatted this way:
613 1 2 README NUL
615 3 1 NUL arch/i386/Makefile NUL arch/x86/Makefile NUL
616 That is:
619 1. the number of added lines;
621 2. a tab;
623 3. the number of deleted lines;
625 4. a tab;
627 5. a NUL (only exists if renamed/copied);
629 6. pathname in preimage;
631 7. a NUL (only exists if renamed/copied);
633 8. pathname in postimage (only exists if renamed/copied);
635 9. a NUL.
637 The extra NUL before the preimage path in renamed case is to allow
639 scripts that read the output to tell if the current record being read
640 is a single-path record or a rename/copy record without reading ahead.
641 After reading added and deleted lines, reading up to NUL would yield
642 the pathname, but if that is NUL, the record will show two paths.
643
645 Various ways to check your working tree
646 $ git diff [1m(1)
648 $ git diff --cached [1m(2)
649 $ git diff HEAD [1m(3)
650 1. Changes in the working tree not yet staged for the next commit.
652 2. Changes between the index and your last commit; what you would
653 be committing if you run "git commit" without "-a" option.
654 3. Changes in the working tree since your last commit; what you
655 would be committing if you run "git commit -a"
656 Comparing with arbitrary commits
658 $ git diff test [1m(1)
660 $ git diff HEAD -- ./test [1m(2)
661 $ git diff HEAD^ HEAD [1m(3)
662 1. Instead of using the tip of the current branch, compare with the
664 tip of "test" branch.
665 2. Instead of comparing with the tip of "test" branch, compare with
666 the tip of the current branch, but limit the comparison to the file
667 "test".
668 3. Compare the version before the last commit and the last commit.
669 Comparing branches
671 $ git diff topic master [1m(1)
673 $ git diff topic..master [1m(2)
674 $ git diff topic...master [1m(3)
675 1. Changes between the tips of the topic and the master branches.
677 2. Same as above.
678 3. Changes that occurred on the master branch since when the topic
679 branch was started off it.
680 Limiting the diff output
682 $ git diff --diff-filter=MRC [1m(1)
684 $ git diff --name-status [1m(2)
685 $ git diff arch/i386 include/asm-i386 [1m(3)
686 1. Show only modification, rename and copy, but not addition nor
688 deletion.
689 2. Show only names and the nature of change, but not actual diff
690 output.
691 3. Limit diff output to named subtrees.
692 Munging the diff output
694 $ git diff --find-copies-harder -B -C [1m(1)
696 $ git diff -R [1m(2)
697 1. Spend extra cycles to find renames, copies and complete rewrites
699 (very expensive).
700 2. Output diff in reverse.
701
703 git-difftool(1)
704 Show changes using common diff tools
705
707 Written by Linus Torvalds <torvalds@osdl.org[1]>
708
710 Documentation by Junio C Hamano and the git-list
711 <git@vger.kernel.org[2]>.
712
714 Part of the git(1) suite
715
717 1. torvalds@osdl.org
718 mailto:torvalds@osdl.org
719 2. git@vger.kernel.org
721 mailto:git@vger.kernel.org
722Git 1.7.1 08/16/2017 GIT-DIFF(1)