1GIT-DIFF-INDEX(1) Git Manual GIT-DIFF-INDEX(1)
2
6 git-diff-index - Compare a tree to the working tree or index
7
9 git diff-index [-m] [--cached] [--merge-base] [<common-diff-options>] <tree-ish> [<path>...]
10
12 Compares the content and mode of the blobs found in a tree object with
13 the corresponding tracked files in the working tree, or with the
14 corresponding paths in the index. When <path> arguments are present,
15 compares only paths matching those patterns. Otherwise all tracked
16 files are compared.
17
19 -p, -u, --patch
20 Generate patch (see section on generating patches).
21 -s, --no-patch
23 Suppress diff output. Useful for commands like git show that show
24 the patch by default, or to cancel the effect of --patch.
25 -U<n>, --unified=<n>
27 Generate diffs with <n> lines of context instead of the usual
28 three. Implies --patch.
29 --output=<file>
31 Output to a specific file instead of stdout.
32 --output-indicator-new=<char>, --output-indicator-old=<char>,
34 --output-indicator-context=<char>
35 Specify the character used to indicate new, old or context lines in
36 the generated patch. Normally they are +, - and ' ' respectively.
37 --raw
39 Generate the diff in raw format. This is the default.
40 --patch-with-raw
42 Synonym for -p --raw.
43 --indent-heuristic
45 Enable the heuristic that shifts diff hunk boundaries to make
46 patches easier to read. This is the default.
47 --no-indent-heuristic
49 Disable the indent heuristic.
50 --minimal
52 Spend extra time to make sure the smallest possible diff is
53 produced.
54 --patience
56 Generate a diff using the "patience diff" algorithm.
57 --histogram
59 Generate a diff using the "histogram diff" algorithm.
60 --anchored=<text>
62 Generate a diff using the "anchored diff" algorithm.
63 This option may be specified more than once.
65 If a line exists in both the source and destination, exists only
67 once, and starts with this text, this algorithm attempts to prevent
68 it from appearing as a deletion or addition in the output. It uses
69 the "patience diff" algorithm internally.
70 --diff-algorithm={patience|minimal|histogram|myers}
72 Choose a diff algorithm. The variants are as follows:
73 default, myers
75 The basic greedy diff algorithm. Currently, this is the
76 default.
77 minimal
79 Spend extra time to make sure the smallest possible diff is
80 produced.
81 patience
83 Use "patience diff" algorithm when generating patches.
84 histogram
86 This algorithm extends the patience algorithm to "support
87 low-occurrence common elements".
88 For instance, if you configured the diff.algorithm variable to a
90 non-default value and want to use the default one, then you have to
91 use --diff-algorithm=default option.
92 --stat[=<width>[,<name-width>[,<count>]]]
94 Generate a diffstat. By default, as much space as necessary will be
95 used for the filename part, and the rest for the graph part.
96 Maximum width defaults to terminal width, or 80 columns if not
97 connected to a terminal, and can be overridden by <width>. The
98 width of the filename part can be limited by giving another width
99 <name-width> after a comma. The width of the graph part can be
100 limited by using --stat-graph-width=<width> (affects all commands
101 generating a stat graph) or by setting diff.statGraphWidth=<width>
102 (does not affect git format-patch). By giving a third parameter
103 <count>, you can limit the output to the first <count> lines,
104 followed by ... if there are more.
105 These parameters can also be set individually with
107 --stat-width=<width>, --stat-name-width=<name-width> and
108 --stat-count=<count>.
109 --compact-summary
111 Output a condensed summary of extended header information such as
112 file creations or deletions ("new" or "gone", optionally "+l" if
113 it’s a symlink) and mode changes ("+x" or "-x" for adding or
114 removing executable bit respectively) in diffstat. The information
115 is put between the filename part and the graph part. Implies
116 --stat.
117 --numstat
119 Similar to --stat, but shows number of added and deleted lines in
120 decimal notation and pathname without abbreviation, to make it more
121 machine friendly. For binary files, outputs two - instead of saying
122 0 0.
123 --shortstat
125 Output only the last line of the --stat format containing total
126 number of modified files, as well as number of added and deleted
127 lines.
128 -X[<param1,param2,...>], --dirstat[=<param1,param2,...>]
130 Output the distribution of relative amount of changes for each
131 sub-directory. The behavior of --dirstat can be customized by
132 passing it a comma separated list of parameters. The defaults are
133 controlled by the diff.dirstat configuration variable (see git-
134 config(1)). The following parameters are available:
135 changes
137 Compute the dirstat numbers by counting the lines that have
138 been removed from the source, or added to the destination. This
139 ignores the amount of pure code movements within a file. In
140 other words, rearranging lines in a file is not counted as much
141 as other changes. This is the default behavior when no
142 parameter is given.
143 lines
145 Compute the dirstat numbers by doing the regular line-based
146 diff analysis, and summing the removed/added line counts. (For
147 binary files, count 64-byte chunks instead, since binary files
148 have no natural concept of lines). This is a more expensive
149 --dirstat behavior than the changes behavior, but it does count
150 rearranged lines within a file as much as other changes. The
151 resulting output is consistent with what you get from the other
152 --*stat options.
153 files
155 Compute the dirstat numbers by counting the number of files
156 changed. Each changed file counts equally in the dirstat
157 analysis. This is the computationally cheapest --dirstat
158 behavior, since it does not have to look at the file contents
159 at all.
160 cumulative
162 Count changes in a child directory for the parent directory as
163 well. Note that when using cumulative, the sum of the
164 percentages reported may exceed 100%. The default
165 (non-cumulative) behavior can be specified with the
166 noncumulative parameter.
167 <limit>
169 An integer parameter specifies a cut-off percent (3% by
170 default). Directories contributing less than this percentage of
171 the changes are not shown in the output.
172 Example: The following will count changed files, while ignoring
174 directories with less than 10% of the total amount of changed
175 files, and accumulating child directory counts in the parent
176 directories: --dirstat=files,10,cumulative.
177 --cumulative
179 Synonym for --dirstat=cumulative
180 --dirstat-by-file[=<param1,param2>...]
182 Synonym for --dirstat=files,param1,param2...
183 --summary
185 Output a condensed summary of extended header information such as
186 creations, renames and mode changes.
187 --patch-with-stat
189 Synonym for -p --stat.
190 -z
192 When --raw, --numstat, --name-only or --name-status has been given,
193 do not munge pathnames and use NULs as output field terminators.
194 Without this option, pathnames with "unusual" characters are quoted
196 as explained for the configuration variable core.quotePath (see
197 git-config(1)).
198 --name-only
200 Show only names of changed files. The file names are often encoded
201 in UTF-8. For more information see the discussion about encoding in
202 the git-log(1) manual page.
203 --name-status
205 Show only names and status of changed files. See the description of
206 the --diff-filter option on what the status letters mean. Just like
207 --name-only the file names are often encoded in UTF-8.
208 --submodule[=<format>]
210 Specify how differences in submodules are shown. When specifying
211 --submodule=short the short format is used. This format just shows
212 the names of the commits at the beginning and end of the range.
213 When --submodule or --submodule=log is specified, the log format is
214 used. This format lists the commits in the range like git-
215 submodule(1) summary does. When --submodule=diff is specified, the
216 diff format is used. This format shows an inline diff of the
217 changes in the submodule contents between the commit range.
218 Defaults to diff.submodule or the short format if the config option
219 is unset.
220 --color[=<when>]
222 Show colored diff. --color (i.e. without =<when>) is the same as
223 --color=always. <when> can be one of always, never, or auto.
224 --no-color
226 Turn off colored diff. It is the same as --color=never.
227 --color-moved[=<mode>]
229 Moved lines of code are colored differently. The <mode> defaults to
230 no if the option is not given and to zebra if the option with no
231 mode is given. The mode must be one of:
232 no
234 Moved lines are not highlighted.
235 default
237 Is a synonym for zebra. This may change to a more sensible mode
238 in the future.
239 plain
241 Any line that is added in one location and was removed in
242 another location will be colored with color.diff.newMoved.
243 Similarly color.diff.oldMoved will be used for removed lines
244 that are added somewhere else in the diff. This mode picks up
245 any moved line, but it is not very useful in a review to
246 determine if a block of code was moved without permutation.
247 blocks
249 Blocks of moved text of at least 20 alphanumeric characters are
250 detected greedily. The detected blocks are painted using either
251 the color.diff.{old,new}Moved color. Adjacent blocks cannot be
252 told apart.
253 zebra
255 Blocks of moved text are detected as in blocks mode. The blocks
256 are painted using either the color.diff.{old,new}Moved color or
257 color.diff.{old,new}MovedAlternative. The change between the
258 two colors indicates that a new block was detected.
259 dimmed-zebra
261 Similar to zebra, but additional dimming of uninteresting parts
262 of moved code is performed. The bordering lines of two adjacent
263 blocks are considered interesting, the rest is uninteresting.
264 dimmed_zebra is a deprecated synonym.
265 --no-color-moved
267 Turn off move detection. This can be used to override configuration
268 settings. It is the same as --color-moved=no.
269 --color-moved-ws=<modes>
271 This configures how whitespace is ignored when performing the move
272 detection for --color-moved. These modes can be given as a comma
273 separated list:
274 no
276 Do not ignore whitespace when performing move detection.
277 ignore-space-at-eol
279 Ignore changes in whitespace at EOL.
280 ignore-space-change
282 Ignore changes in amount of whitespace. This ignores whitespace
283 at line end, and considers all other sequences of one or more
284 whitespace characters to be equivalent.
285 ignore-all-space
287 Ignore whitespace when comparing lines. This ignores
288 differences even if one line has whitespace where the other
289 line has none.
290 allow-indentation-change
292 Initially ignore any whitespace in the move detection, then
293 group the moved code blocks only into a block if the change in
294 whitespace is the same per line. This is incompatible with the
295 other modes.
296 --no-color-moved-ws
298 Do not ignore whitespace when performing move detection. This can
299 be used to override configuration settings. It is the same as
300 --color-moved-ws=no.
301 --word-diff[=<mode>]
303 Show a word diff, using the <mode> to delimit changed words. By
304 default, words are delimited by whitespace; see --word-diff-regex
305 below. The <mode> defaults to plain, and must be one of:
306 color
308 Highlight changed words using only colors. Implies --color.
309 plain
311 Show words as [-removed-] and {+added+}. Makes no attempts to
312 escape the delimiters if they appear in the input, so the
313 output may be ambiguous.
314 porcelain
316 Use a special line-based format intended for script
317 consumption. Added/removed/unchanged runs are printed in the
318 usual unified diff format, starting with a +/-/` ` character at
319 the beginning of the line and extending to the end of the line.
320 Newlines in the input are represented by a tilde ~ on a line of
321 its own.
322 none
324 Disable word diff again.
325 Note that despite the name of the first mode, color is used to
327 highlight the changed parts in all modes if enabled.
328 --word-diff-regex=<regex>
330 Use <regex> to decide what a word is, instead of considering runs
331 of non-whitespace to be a word. Also implies --word-diff unless it
332 was already enabled.
333 Every non-overlapping match of the <regex> is considered a word.
335 Anything between these matches is considered whitespace and
336 ignored(!) for the purposes of finding differences. You may want to
337 append |[^[:space:]] to your regular expression to make sure that
338 it matches all non-whitespace characters. A match that contains a
339 newline is silently truncated(!) at the newline.
340 For example, --word-diff-regex=. will treat each character as a
342 word and, correspondingly, show differences character by character.
343 The regex can also be set via a diff driver or configuration
345 option, see gitattributes(5) or git-config(1). Giving it explicitly
346 overrides any diff driver or configuration setting. Diff drivers
347 override configuration settings.
348 --color-words[=<regex>]
350 Equivalent to --word-diff=color plus (if a regex was specified)
351 --word-diff-regex=<regex>.
352 --no-renames
354 Turn off rename detection, even when the configuration file gives
355 the default to do so.
356 --[no-]rename-empty
358 Whether to use empty blobs as rename source.
359 --check
361 Warn if changes introduce conflict markers or whitespace errors.
362 What are considered whitespace errors is controlled by
363 core.whitespace configuration. By default, trailing whitespaces
364 (including lines that consist solely of whitespaces) and a space
365 character that is immediately followed by a tab character inside
366 the initial indent of the line are considered whitespace errors.
367 Exits with non-zero status if problems are found. Not compatible
368 with --exit-code.
369 --ws-error-highlight=<kind>
371 Highlight whitespace errors in the context, old or new lines of the
372 diff. Multiple values are separated by comma, none resets previous
373 values, default reset the list to new and all is a shorthand for
374 old,new,context. When this option is not given, and the
375 configuration variable diff.wsErrorHighlight is not set, only
376 whitespace errors in new lines are highlighted. The whitespace
377 errors are colored with color.diff.whitespace.
378 --full-index
380 Instead of the first handful of characters, show the full pre- and
381 post-image blob object names on the "index" line when generating
382 patch format output.
383 --binary
385 In addition to --full-index, output a binary diff that can be
386 applied with git-apply. Implies --patch.
387 --abbrev[=<n>]
389 Instead of showing the full 40-byte hexadecimal object name in
390 diff-raw format output and diff-tree header lines, show the
391 shortest prefix that is at least <n> hexdigits long that uniquely
392 refers the object. In diff-patch output format, --full-index takes
393 higher precedence, i.e. if --full-index is specified, full blob
394 names will be shown regardless of --abbrev. Non default number of
395 digits can be specified with --abbrev=<n>.
396 -B[<n>][/<m>], --break-rewrites[=[<n>][/<m>]]
398 Break complete rewrite changes into pairs of delete and create.
399 This serves two purposes:
400 It affects the way a change that amounts to a total rewrite of a
402 file not as a series of deletion and insertion mixed together with
403 a very few lines that happen to match textually as the context, but
404 as a single deletion of everything old followed by a single
405 insertion of everything new, and the number m controls this aspect
406 of the -B option (defaults to 60%). -B/70% specifies that less
407 than 30% of the original should remain in the result for Git to
408 consider it a total rewrite (i.e. otherwise the resulting patch
409 will be a series of deletion and insertion mixed together with
410 context lines).
411 When used with -M, a totally-rewritten file is also considered as
413 the source of a rename (usually -M only considers a file that
414 disappeared as the source of a rename), and the number n controls
415 this aspect of the -B option (defaults to 50%). -B20% specifies
416 that a change with addition and deletion compared to 20% or more of
417 the file’s size are eligible for being picked up as a possible
418 source of a rename to another file.
419 -M[<n>], --find-renames[=<n>]
421 Detect renames. If n is specified, it is a threshold on the
422 similarity index (i.e. amount of addition/deletions compared to the
423 file’s size). For example, -M90% means Git should consider a
424 delete/add pair to be a rename if more than 90% of the file hasn’t
425 changed. Without a % sign, the number is to be read as a fraction,
426 with a decimal point before it. I.e., -M5 becomes 0.5, and is thus
427 the same as -M50%. Similarly, -M05 is the same as -M5%. To limit
428 detection to exact renames, use -M100%. The default similarity
429 index is 50%.
430 -C[<n>], --find-copies[=<n>]
432 Detect copies as well as renames. See also --find-copies-harder. If
433 n is specified, it has the same meaning as for -M<n>.
434 --find-copies-harder
436 For performance reasons, by default, -C option finds copies only if
437 the original file of the copy was modified in the same changeset.
438 This flag makes the command inspect unmodified files as candidates
439 for the source of copy. This is a very expensive operation for
440 large projects, so use it with caution. Giving more than one -C
441 option has the same effect.
442 -D, --irreversible-delete
444 Omit the preimage for deletes, i.e. print only the header but not
445 the diff between the preimage and /dev/null. The resulting patch is
446 not meant to be applied with patch or git apply; this is solely for
447 people who want to just concentrate on reviewing the text after the
448 change. In addition, the output obviously lacks enough information
449 to apply such a patch in reverse, even manually, hence the name of
450 the option.
451 When used together with -B, omit also the preimage in the deletion
453 part of a delete/create pair.
454 -l<num>
456 The -M and -C options involve some preliminary steps that can
457 detect subsets of renames/copies cheaply, followed by an exhaustive
458 fallback portion that compares all remaining unpaired destinations
459 to all relevant sources. (For renames, only remaining unpaired
460 sources are relevant; for copies, all original sources are
461 relevant.) For N sources and destinations, this exhaustive check is
462 O(N^2). This option prevents the exhaustive portion of rename/copy
463 detection from running if the number of source/destination files
464 involved exceeds the specified number. Defaults to
465 diff.renameLimit. Note that a value of 0 is treated as unlimited.
466 --diff-filter=[(A|C|D|M|R|T|U|X|B)...[*]]
468 Select only files that are Added (A), Copied (C), Deleted (D),
469 Modified (M), Renamed (R), have their type (i.e. regular file,
470 symlink, submodule, ...) changed (T), are Unmerged (U), are Unknown
471 (X), or have had their pairing Broken (B). Any combination of the
472 filter characters (including none) can be used. When *
473 (All-or-none) is added to the combination, all paths are selected
474 if there is any file that matches other criteria in the comparison;
475 if there is no file that matches other criteria, nothing is
476 selected.
477 Also, these upper-case letters can be downcased to exclude. E.g.
479 --diff-filter=ad excludes added and deleted paths.
480 Note that not all diffs can feature all types. For instance, copied
482 and renamed entries cannot appear if detection for those types is
483 disabled.
484 -S<string>
486 Look for differences that change the number of occurrences of the
487 specified string (i.e. addition/deletion) in a file. Intended for
488 the scripter’s use.
489 It is useful when you’re looking for an exact block of code (like a
491 struct), and want to know the history of that block since it first
492 came into being: use the feature iteratively to feed the
493 interesting block in the preimage back into -S, and keep going
494 until you get the very first version of the block.
495 Binary files are searched as well.
497 -G<regex>
499 Look for differences whose patch text contains added/removed lines
500 that match <regex>.
501 To illustrate the difference between -S<regex> --pickaxe-regex and
503 -G<regex>, consider a commit with the following diff in the same
504 file:
505 + return frotz(nitfol, two->ptr, 1, 0);
507 ...
508 - hit = frotz(nitfol, mf2.ptr, 1, 0);
509 While git log -G"frotz\(nitfol" will show this commit, git log
511 -S"frotz\(nitfol" --pickaxe-regex will not (because the number of
512 occurrences of that string did not change).
513 Unless --text is supplied patches of binary files without a
515 textconv filter will be ignored.
516 See the pickaxe entry in gitdiffcore(7) for more information.
518 --find-object=<object-id>
520 Look for differences that change the number of occurrences of the
521 specified object. Similar to -S, just the argument is different in
522 that it doesn’t search for a specific string but for a specific
523 object id.
524 The object can be a blob or a submodule commit. It implies the -t
526 option in git-log to also find trees.
527 --pickaxe-all
529 When -S or -G finds a change, show all the changes in that
530 changeset, not just the files that contain the change in <string>.
531 --pickaxe-regex
533 Treat the <string> given to -S as an extended POSIX regular
534 expression to match.
535 -O<orderfile>
537 Control the order in which files appear in the output. This
538 overrides the diff.orderFile configuration variable (see git-
539 config(1)). To cancel diff.orderFile, use -O/dev/null.
540 The output order is determined by the order of glob patterns in
542 <orderfile>. All files with pathnames that match the first pattern
543 are output first, all files with pathnames that match the second
544 pattern (but not the first) are output next, and so on. All files
545 with pathnames that do not match any pattern are output last, as if
546 there was an implicit match-all pattern at the end of the file. If
547 multiple pathnames have the same rank (they match the same pattern
548 but no earlier patterns), their output order relative to each other
549 is the normal order.
550 <orderfile> is parsed as follows:
552 • Blank lines are ignored, so they can be used as separators for
554 readability.
555 • Lines starting with a hash ("#") are ignored, so they can be
557 used for comments. Add a backslash ("\") to the beginning of
558 the pattern if it starts with a hash.
559 • Each other line contains a single pattern.
561 Patterns have the same syntax and semantics as patterns used for
563 fnmatch(3) without the FNM_PATHNAME flag, except a pathname also
564 matches a pattern if removing any number of the final pathname
565 components matches the pattern. For example, the pattern "foo*bar"
566 matches "fooasdfbar" and "foo/bar/baz/asdf" but not "foobarx".
567 --skip-to=<file>, --rotate-to=<file>
569 Discard the files before the named <file> from the output (i.e.
570 skip to), or move them to the end of the output (i.e. rotate to).
571 These were invented primarily for use of the git difftool command,
572 and may not be very useful otherwise.
573 -R
575 Swap two inputs; that is, show differences from index or on-disk
576 file to tree contents.
577 --relative[=<path>], --no-relative
579 When run from a subdirectory of the project, it can be told to
580 exclude changes outside the directory and show pathnames relative
581 to it with this option. When you are not in a subdirectory (e.g. in
582 a bare repository), you can name which subdirectory to make the
583 output relative to by giving a <path> as an argument.
584 --no-relative can be used to countermand both diff.relative config
585 option and previous --relative.
586 -a, --text
588 Treat all files as text.
589 --ignore-cr-at-eol
591 Ignore carriage-return at the end of line when doing a comparison.
592 --ignore-space-at-eol
594 Ignore changes in whitespace at EOL.
595 -b, --ignore-space-change
597 Ignore changes in amount of whitespace. This ignores whitespace at
598 line end, and considers all other sequences of one or more
599 whitespace characters to be equivalent.
600 -w, --ignore-all-space
602 Ignore whitespace when comparing lines. This ignores differences
603 even if one line has whitespace where the other line has none.
604 --ignore-blank-lines
606 Ignore changes whose lines are all blank.
607 -I<regex>, --ignore-matching-lines=<regex>
609 Ignore changes whose all lines match <regex>. This option may be
610 specified more than once.
611 --inter-hunk-context=<lines>
613 Show the context between diff hunks, up to the specified number of
614 lines, thereby fusing hunks that are close to each other. Defaults
615 to diff.interHunkContext or 0 if the config option is unset.
616 -W, --function-context
618 Show whole function as context lines for each change. The function
619 names are determined in the same way as git diff works out patch
620 hunk headers (see Defining a custom hunk-header in
621 gitattributes(5)).
622 --exit-code
624 Make the program exit with codes similar to diff(1). That is, it
625 exits with 1 if there were differences and 0 means no differences.
626 --quiet
628 Disable all output of the program. Implies --exit-code.
629 --ext-diff
631 Allow an external diff helper to be executed. If you set an
632 external diff driver with gitattributes(5), you need to use this
633 option with git-log(1) and friends.
634 --no-ext-diff
636 Disallow external diff drivers.
637 --textconv, --no-textconv
639 Allow (or disallow) external text conversion filters to be run when
640 comparing binary files. See gitattributes(5) for details. Because
641 textconv filters are typically a one-way conversion, the resulting
642 diff is suitable for human consumption, but cannot be applied. For
643 this reason, textconv filters are enabled by default only for git-
644 diff(1) and git-log(1), but not for git-format-patch(1) or diff
645 plumbing commands.
646 --ignore-submodules[=<when>]
648 Ignore changes to submodules in the diff generation. <when> can be
649 either "none", "untracked", "dirty" or "all", which is the default.
650 Using "none" will consider the submodule modified when it either
651 contains untracked or modified files or its HEAD differs from the
652 commit recorded in the superproject and can be used to override any
653 settings of the ignore option in git-config(1) or gitmodules(5).
654 When "untracked" is used submodules are not considered dirty when
655 they only contain untracked content (but they are still scanned for
656 modified content). Using "dirty" ignores all changes to the work
657 tree of submodules, only changes to the commits stored in the
658 superproject are shown (this was the behavior until 1.7.0). Using
659 "all" hides all changes to submodules.
660 --src-prefix=<prefix>
662 Show the given source prefix instead of "a/".
663 --dst-prefix=<prefix>
665 Show the given destination prefix instead of "b/".
666 --no-prefix
668 Do not show any source or destination prefix.
669 --line-prefix=<prefix>
671 Prepend an additional prefix to every line of output.
672 --ita-invisible-in-index
674 By default entries added by "git add -N" appear as an existing
675 empty file in "git diff" and a new file in "git diff --cached".
676 This option makes the entry appear as a new file in "git diff" and
677 non-existent in "git diff --cached". This option could be reverted
678 with --ita-visible-in-index. Both options are experimental and
679 could be removed in future.
680 For more detailed explanation on these common options, see also
682 gitdiffcore(7).
683 <tree-ish>
685 The id of a tree object to diff against.
686 --cached
688 Do not consider the on-disk file at all.
689 --merge-base
691 Instead of comparing <tree-ish> directly, use the merge base
692 between <tree-ish> and HEAD instead. <tree-ish> must be a commit.
693 -m
695 By default, files recorded in the index but not checked out are
696 reported as deleted. This flag makes git diff-index say that all
697 non-checked-out files are up to date.
698
700 The raw output format from "git-diff-index", "git-diff-tree",
701 "git-diff-files" and "git diff --raw" are very similar.
702 These commands all compare two sets of things; what is compared
704 differs:
705 git-diff-index <tree-ish>
707 compares the <tree-ish> and the files on the filesystem.
708 git-diff-index --cached <tree-ish>
710 compares the <tree-ish> and the index.
711 git-diff-tree [-r] <tree-ish-1> <tree-ish-2> [<pattern>...]
713 compares the trees named by the two arguments.
714 git-diff-files [<pattern>...]
716 compares the index and the files on the filesystem.
717 The "git-diff-tree" command begins its output by printing the hash of
719 what is being compared. After that, all the commands print one output
720 line per changed file.
721 An output line is formatted this way:
723 in-place edit :100644 100644 bcd1234 0123456 M file0
725 copy-edit :100644 100644 abcd123 1234567 C68 file1 file2
726 rename-edit :100644 100644 abcd123 1234567 R86 file1 file3
727 create :000000 100644 0000000 1234567 A file4
728 delete :100644 000000 1234567 0000000 D file5
729 unmerged :000000 000000 0000000 0000000 U file6
730 That is, from the left to the right:
732 1. a colon.
734 2. mode for "src"; 000000 if creation or unmerged.
736 3. a space.
738 4. mode for "dst"; 000000 if deletion or unmerged.
740 5. a space.
742 6. sha1 for "src"; 0{40} if creation or unmerged.
744 7. a space.
746 8. sha1 for "dst"; 0{40} if creation, unmerged or "look at work tree".
748 9. a space.
750 10. status, followed by optional "score" number.
752 11. a tab or a NUL when -z option is used.
754 12. path for "src"
756 13. a tab or a NUL when -z option is used; only exists for C or R.
758 14. path for "dst"; only exists for C or R.
760 15. an LF or a NUL when -z option is used, to terminate the record.
762 Possible status letters are:
764 • A: addition of a file
766 • C: copy of a file into a new one
768 • D: deletion of a file
770 • M: modification of the contents or mode of a file
772 • R: renaming of a file
774 • T: change in the type of the file (regular file, symbolic link or
776 submodule)
777 • U: file is unmerged (you must complete the merge before it can be
779 committed)
780 • X: "unknown" change type (most probably a bug, please report it)
782 Status letters C and R are always followed by a score (denoting the
784 percentage of similarity between the source and target of the move or
785 copy). Status letter M may be followed by a score (denoting the
786 percentage of dissimilarity) for file rewrites.
787 <sha1> is shown as all 0’s if a file is new on the filesystem and it is
789 out of sync with the index.
790 Example:
792 :100644 100644 5be4a4a 0000000 M file.c
794 Without the -z option, pathnames with "unusual" characters are quoted
796 as explained for the configuration variable core.quotePath (see git-
797 config(1)). Using -z the filename is output verbatim and the line is
798 terminated by a NUL byte.
799
801 "git-diff-tree", "git-diff-files" and "git-diff --raw" can take -c or
802 --cc option to generate diff output also for merge commits. The output
803 differs from the format described above in the following way:
804 1. there is a colon for each parent
806 2. there are more "src" modes and "src" sha1
808 3. status is concatenated status characters for each parent
810 4. no optional "score" number
812 5. tab-separated pathname(s) of the file
814 For -c and --cc, only the destination or final path is shown even if
816 the file was renamed on any side of history. With --combined-all-paths,
817 the name of the path in each parent is shown followed by the name of
818 the path in the merge commit.
819 Examples for -c and --cc without --combined-all-paths:
821 ::100644 100644 100644 fabadb8 cc95eb0 4866510 MM desc.c
823 ::100755 100755 100755 52b7a2d 6d1ac04 d2ac7d7 RM bar.sh
824 ::100644 100644 100644 e07d6c5 9042e82 ee91881 RR phooey.c
825 Examples when --combined-all-paths added to either -c or --cc:
827 ::100644 100644 100644 fabadb8 cc95eb0 4866510 MM desc.c desc.c desc.c
829 ::100755 100755 100755 52b7a2d 6d1ac04 d2ac7d7 RM foo.sh bar.sh bar.sh
830 ::100644 100644 100644 e07d6c5 9042e82 ee91881 RR fooey.c fuey.c phooey.c
831 Note that combined diff lists only files which were modified from all
833 parents.
834
836 Running git-diff(1), git-log(1), git-show(1), git-diff-index(1), git-
837 diff-tree(1), or git-diff-files(1) with the -p option produces patch
838 text. You can customize the creation of patch text via the
839 GIT_EXTERNAL_DIFF and the GIT_DIFF_OPTS environment variables (see
840 git(1)), and the diff attribute (see gitattributes(5)).
841 What the -p option produces is slightly different from the traditional
843 diff format:
844 1. It is preceded with a "git diff" header that looks like this:
846 diff --git a/file1 b/file2
848 The a/ and b/ filenames are the same unless rename/copy is
850 involved. Especially, even for a creation or a deletion, /dev/null
851 is not used in place of the a/ or b/ filenames.
852 When rename/copy is involved, file1 and file2 show the name of the
854 source file of the rename/copy and the name of the file that
855 rename/copy produces, respectively.
856 2. It is followed by one or more extended header lines:
858 old mode <mode>
860 new mode <mode>
861 deleted file mode <mode>
862 new file mode <mode>
863 copy from <path>
864 copy to <path>
865 rename from <path>
866 rename to <path>
867 similarity index <number>
868 dissimilarity index <number>
869 index <hash>..<hash> <mode>
870 File modes are printed as 6-digit octal numbers including the file
872 type and file permission bits.
873 Path names in extended headers do not include the a/ and b/
875 prefixes.
876 The similarity index is the percentage of unchanged lines, and the
878 dissimilarity index is the percentage of changed lines. It is a
879 rounded down integer, followed by a percent sign. The similarity
880 index value of 100% is thus reserved for two equal files, while
881 100% dissimilarity means that no line from the old file made it
882 into the new one.
883 The index line includes the blob object names before and after the
885 change. The <mode> is included if the file mode does not change;
886 otherwise, separate lines indicate the old and the new mode.
887 3. Pathnames with "unusual" characters are quoted as explained for the
889 configuration variable core.quotePath (see git-config(1)).
890 4. All the file1 files in the output refer to files before the commit,
892 and all the file2 files refer to files after the commit. It is
893 incorrect to apply each change to each file sequentially. For
894 example, this patch will swap a and b:
895 diff --git a/a b/b
897 rename from a
898 rename to b
899 diff --git a/b b/a
900 rename from b
901 rename to a
902 5. Hunk headers mention the name of the function to which the hunk
904 applies. See "Defining a custom hunk-header" in gitattributes(5)
905 for details of how to tailor to this to specific languages.
906
908 Any diff-generating command can take the -c or --cc option to produce a
909 combined diff when showing a merge. This is the default format when
910 showing merges with git-diff(1) or git-show(1). Note also that you can
911 give suitable --diff-merges option to any of these commands to force
912 generation of diffs in specific format.
913 A "combined diff" format looks like this:
915 diff --combined describe.c
917 index fabadb8,cc95eb0..4866510
918 --- a/describe.c
919 +++ b/describe.c
920 @@@ -98,20 -98,12 +98,20 @@@
921 return (a_date > b_date) ? -1 : (a_date == b_date) ? 0 : 1;
922 }
923 - static void describe(char *arg)
925 -static void describe(struct commit *cmit, int last_one)
926 ++static void describe(char *arg, int last_one)
927 {
928 + unsigned char sha1[20];
929 + struct commit *cmit;
930 struct commit_list *list;
931 static int initialized = 0;
932 struct commit_name *n;
933 + if (get_sha1(arg, sha1) < 0)
935 + usage(describe_usage);
936 + cmit = lookup_commit_reference(sha1);
937 + if (!cmit)
938 + usage(describe_usage);
939 +
940 if (!initialized) {
941 initialized = 1;
942 for_each_ref(get_name);
943 1. It is preceded with a "git diff" header, that looks like this (when
945 the -c option is used):
946 diff --combined file
948 or like this (when the --cc option is used):
950 diff --cc file
952 2. It is followed by one or more extended header lines (this example
954 shows a merge with two parents):
955 index <hash>,<hash>..<hash>
957 mode <mode>,<mode>..<mode>
958 new file mode <mode>
959 deleted file mode <mode>,<mode>
960 The mode <mode>,<mode>..<mode> line appears only if at least one of
962 the <mode> is different from the rest. Extended headers with
963 information about detected contents movement (renames and copying
964 detection) are designed to work with diff of two <tree-ish> and are
965 not used by combined diff format.
966 3. It is followed by two-line from-file/to-file header
968 --- a/file
970 +++ b/file
971 Similar to two-line header for traditional unified diff format,
973 /dev/null is used to signal created or deleted files.
974 However, if the --combined-all-paths option is provided, instead of
976 a two-line from-file/to-file you get a N+1 line from-file/to-file
977 header, where N is the number of parents in the merge commit
978 --- a/file
980 --- a/file
981 --- a/file
982 +++ b/file
983 This extended format can be useful if rename or copy detection is
985 active, to allow you to see the original name of the file in
986 different parents.
987 4. Chunk header format is modified to prevent people from accidentally
989 feeding it to patch -p1. Combined diff format was created for
990 review of merge commit changes, and was not meant to be applied.
991 The change is similar to the change in the extended index header:
992 @@@ <from-file-range> <from-file-range> <to-file-range> @@@
994 There are (number of parents + 1) @ characters in the chunk header
996 for combined diff format.
997 Unlike the traditional unified diff format, which shows two files A and
999 B with a single column that has - (minus — appears in A but removed in
1000 B), + (plus — missing in A but added to B), or " " (space — unchanged)
1001 prefix, this format compares two or more files file1, file2,... with
1002 one file X, and shows how X differs from each of fileN. One column for
1003 each of fileN is prepended to the output line to note how X’s line is
1004 different from it.
1005 A - character in the column N means that the line appears in fileN but
1007 it does not appear in the result. A + character in the column N means
1008 that the line appears in the result, and fileN does not have that line
1009 (in other words, the line was added, from the point of view of that
1010 parent).
1011 In the above example output, the function signature was changed from
1013 both files (hence two - removals from both file1 and file2, plus ++ to
1014 mean one line that was added does not appear in either file1 or file2).
1015 Also eight other lines are the same from file1 but do not appear in
1016 file2 (hence prefixed with +).
1017 When shown by git diff-tree -c, it compares the parents of a merge
1019 commit with the merge result (i.e. file1..fileN are the parents). When
1020 shown by git diff-files -c, it compares the two unresolved merge
1021 parents with the working tree file (i.e. file1 is stage 2 aka "our
1022 version", file2 is stage 3 aka "their version").
1023
1025 The --summary option describes newly added, deleted, renamed and copied
1026 files. The --stat option adds diffstat(1) graph to the output. These
1027 options can be combined with other options, such as -p, and are meant
1028 for human consumption.
1029 When showing a change that involves a rename or a copy, --stat output
1031 formats the pathnames compactly by combining common prefix and suffix
1032 of the pathnames. For example, a change that moves arch/i386/Makefile
1033 to arch/x86/Makefile while modifying 4 lines will be shown like this:
1034 arch/{i386 => x86}/Makefile | 4 +--
1036 The --numstat option gives the diffstat(1) information but is designed
1038 for easier machine consumption. An entry in --numstat output looks like
1039 this:
1040 1 2 README
1042 3 1 arch/{i386 => x86}/Makefile
1043 That is, from left to right:
1045 1. the number of added lines;
1047 2. a tab;
1049 3. the number of deleted lines;
1051 4. a tab;
1053 5. pathname (possibly with rename/copy information);
1055 6. a newline.
1057 When -z output option is in effect, the output is formatted this way:
1059 1 2 README NUL
1061 3 1 NUL arch/i386/Makefile NUL arch/x86/Makefile NUL
1062 That is:
1064 1. the number of added lines;
1066 2. a tab;
1068 3. the number of deleted lines;
1070 4. a tab;
1072 5. a NUL (only exists if renamed/copied);
1074 6. pathname in preimage;
1076 7. a NUL (only exists if renamed/copied);
1078 8. pathname in postimage (only exists if renamed/copied);
1080 9. a NUL.
1082 The extra NUL before the preimage path in renamed case is to allow
1084 scripts that read the output to tell if the current record being read
1085 is a single-path record or a rename/copy record without reading ahead.
1086 After reading added and deleted lines, reading up to NUL would yield
1087 the pathname, but if that is NUL, the record will show two paths.
1088
1090 You can choose whether you want to trust the index file entirely (using
1091 the --cached flag) or ask the diff logic to show any files that don’t
1092 match the stat state as being "tentatively changed". Both of these
1093 operations are very useful indeed.
1094
1096 If --cached is specified, it allows you to ask:
1097 show me the differences between HEAD and the current index
1099 contents (the ones I'd write using 'git write-tree')
1100 For example, let’s say that you have worked on your working directory,
1102 updated some files in the index and are ready to commit. You want to
1103 see exactly what you are going to commit, without having to write a new
1104 tree object and compare it that way, and to do that, you just do
1105 git diff-index --cached HEAD
1107 Example: let’s say I had renamed commit.c to git-commit.c, and I had
1109 done an update-index to make that effective in the index file. git
1110 diff-files wouldn’t show anything at all, since the index file matches
1111 my working directory. But doing a git diff-index does:
1112 torvalds@ppc970:~/git> git diff-index --cached HEAD
1114 -100644 blob 4161aecc6700a2eb579e842af0b7f22b98443f74 commit.c
1115 +100644 blob 4161aecc6700a2eb579e842af0b7f22b98443f74 git-commit.c
1116 You can see easily that the above is a rename.
1118 In fact, git diff-index --cached should always be entirely equivalent
1120 to actually doing a git write-tree and comparing that. Except this one
1121 is much nicer for the case where you just want to check where you are.
1122 So doing a git diff-index --cached is basically very useful when you
1124 are asking yourself "what have I already marked for being committed,
1125 and what’s the difference to a previous tree".
1126
1128 The "non-cached" mode takes a different approach, and is potentially
1129 the more useful of the two in that what it does can’t be emulated with
1130 a git write-tree + git diff-tree. Thus that’s the default mode. The
1131 non-cached version asks the question:
1132 show me the differences between HEAD and the currently checked out
1134 tree - index contents _and_ files that aren't up to date
1135 which is obviously a very useful question too, since that tells you
1137 what you could commit. Again, the output matches the git diff-tree -r
1138 output to a tee, but with a twist.
1139 The twist is that if some file doesn’t match the index, we don’t have a
1141 backing store thing for it, and we use the magic "all-zero" sha1 to
1142 show that. So let’s say that you have edited kernel/sched.c, but have
1143 not actually done a git update-index on it yet - there is no "object"
1144 associated with the new state, and you get:
1145 torvalds@ppc970:~/v2.6/linux> git diff-index --abbrev HEAD
1147 :100644 100664 7476bb... 000000... kernel/sched.c
1148 i.e., it shows that the tree has changed, and that kernel/sched.c is
1150 not up to date and may contain new stuff. The all-zero sha1 means that
1151 to get the real diff, you need to look at the object in the working
1152 directory directly rather than do an object-to-object diff.
1153 Note
1155 As with other commands of this type, git diff-index does not
1156 actually look at the contents of the file at all. So maybe
1157 kernel/sched.c hasn’t actually changed, and it’s just that you
1158 touched it. In either case, it’s a note that you need to git
1159 update-index it to make the index be in sync.
1160 Note
1162 You can have a mixture of files show up as "has been updated" and
1163 "is still dirty in the working directory" together. You can always
1164 tell which file is in which state, since the "has been updated"
1165 ones show a valid sha1, and the "not in sync with the index" ones
1166 will always have the special all-zero sha1.
1167
1169 Part of the git(1) suite
1170Git 2.36.1 2022-05-05 GIT-DIFF-INDEX(1)