1GIT-FORMAT-PATCH(1) Git Manual GIT-FORMAT-PATCH(1)
2
6 git-format-patch - Prepare patches for e-mail submission
7
9 git format-patch [-k] [(-o|--output-directory) <dir> | --stdout]
10 [--no-thread | --thread[=<style>]]
11 [(--attach|--inline)[=<boundary>] | --no-attach]
12 [-s | --signoff]
13 [--signature=<signature> | --no-signature]
14 [--signature-file=<file>]
15 [-n | --numbered | -N | --no-numbered]
16 [--start-number <n>] [--numbered-files]
17 [--in-reply-to=<message id>] [--suffix=.<sfx>]
18 [--ignore-if-in-upstream]
19 [--cover-from-description=<mode>]
20 [--rfc] [--subject-prefix=<subject prefix>]
21 [(--reroll-count|-v) <n>]
22 [--to=<email>] [--cc=<email>]
23 [--[no-]cover-letter] [--quiet]
24 [--[no-]encode-email-headers]
25 [--no-notes | --notes[=<ref>]]
26 [--interdiff=<previous>]
27 [--range-diff=<previous> [--creation-factor=<percent>]]
28 [--filename-max-length=<n>]
29 [--progress]
30 [<common diff options>]
31 [ <since> | <revision range> ]
32
34 Prepare each non-merge commit with its "patch" in one "message" per
35 commit, formatted to resemble a UNIX mailbox. The output of this
36 command is convenient for e-mail submission or for use with git am.
37 A "message" generated by the command consists of three parts:
39 • A brief metadata header that begins with From <commit> with a fixed
41 Mon Sep 17 00:00:00 2001 datestamp to help programs like "file(1)"
42 to recognize that the file is an output from this command, fields
43 that record the author identity, the author date, and the title of
44 the change (taken from the first paragraph of the commit log
45 message).
46 • The second and subsequent paragraphs of the commit log message.
48 • The "patch", which is the "diff -p --stat" output (see git-diff(1))
50 between the commit and its parent.
51 The log message and the patch is separated by a line with a three-dash
53 line.
54 There are two ways to specify which commits to operate on.
56 1. A single commit, <since>, specifies that the commits leading to the
58 tip of the current branch that are not in the history that leads to
59 the <since> to be output.
60 2. Generic <revision range> expression (see "SPECIFYING REVISIONS"
62 section in gitrevisions(7)) means the commits in the specified
63 range.
64 The first rule takes precedence in the case of a single <commit>. To
66 apply the second rule, i.e., format everything since the beginning of
67 history up until <commit>, use the --root option: git format-patch
68 --root <commit>. If you want to format only <commit> itself, you can do
69 this with git format-patch -1 <commit>.
70 By default, each output file is numbered sequentially from 1, and uses
72 the first line of the commit message (massaged for pathname safety) as
73 the filename. With the --numbered-files option, the output file names
74 will only be numbers, without the first line of the commit appended.
75 The names of the output files are printed to standard output, unless
76 the --stdout option is specified.
77 If -o is specified, output files are created in <dir>. Otherwise they
79 are created in the current working directory. The default path can be
80 set with the format.outputDirectory configuration option. The -o option
81 takes precedence over format.outputDirectory. To store patches in the
82 current working directory even when format.outputDirectory points
83 elsewhere, use -o .. All directory components will be created.
84 By default, the subject of a single patch is "[PATCH] " followed by the
86 concatenation of lines from the commit message up to the first blank
87 line (see the DISCUSSION section of git-commit(1)).
88 When multiple patches are output, the subject prefix will instead be
90 "[PATCH n/m] ". To force 1/1 to be added for a single patch, use -n. To
91 omit patch numbers from the subject, use -N.
92 If given --thread, git-format-patch will generate In-Reply-To and
94 References headers to make the second and subsequent patch mails appear
95 as replies to the first mail; this also generates a Message-Id header
96 to reference.
97
99 -p, --no-stat
100 Generate plain patches without any diffstats.
101 -U<n>, --unified=<n>
103 Generate diffs with <n> lines of context instead of the usual
104 three.
105 --output=<file>
107 Output to a specific file instead of stdout.
108 --output-indicator-new=<char>, --output-indicator-old=<char>,
110 --output-indicator-context=<char>
111 Specify the character used to indicate new, old or context lines in
112 the generated patch. Normally they are +, - and ' ' respectively.
113 --indent-heuristic
115 Enable the heuristic that shifts diff hunk boundaries to make
116 patches easier to read. This is the default.
117 --no-indent-heuristic
119 Disable the indent heuristic.
120 --minimal
122 Spend extra time to make sure the smallest possible diff is
123 produced.
124 --patience
126 Generate a diff using the "patience diff" algorithm.
127 --histogram
129 Generate a diff using the "histogram diff" algorithm.
130 --anchored=<text>
132 Generate a diff using the "anchored diff" algorithm.
133 This option may be specified more than once.
135 If a line exists in both the source and destination, exists only
137 once, and starts with this text, this algorithm attempts to prevent
138 it from appearing as a deletion or addition in the output. It uses
139 the "patience diff" algorithm internally.
140 --diff-algorithm={patience|minimal|histogram|myers}
142 Choose a diff algorithm. The variants are as follows:
143 default, myers
145 The basic greedy diff algorithm. Currently, this is the
146 default.
147 minimal
149 Spend extra time to make sure the smallest possible diff is
150 produced.
151 patience
153 Use "patience diff" algorithm when generating patches.
154 histogram
156 This algorithm extends the patience algorithm to "support
157 low-occurrence common elements".
158 For instance, if you configured the diff.algorithm variable to a
160 non-default value and want to use the default one, then you have to
161 use --diff-algorithm=default option.
162 --stat[=<width>[,<name-width>[,<count>]]]
164 Generate a diffstat. By default, as much space as necessary will be
165 used for the filename part, and the rest for the graph part.
166 Maximum width defaults to terminal width, or 80 columns if not
167 connected to a terminal, and can be overridden by <width>. The
168 width of the filename part can be limited by giving another width
169 <name-width> after a comma. The width of the graph part can be
170 limited by using --stat-graph-width=<width> (affects all commands
171 generating a stat graph) or by setting diff.statGraphWidth=<width>
172 (does not affect git format-patch). By giving a third parameter
173 <count>, you can limit the output to the first <count> lines,
174 followed by ... if there are more.
175 These parameters can also be set individually with
177 --stat-width=<width>, --stat-name-width=<name-width> and
178 --stat-count=<count>.
179 --compact-summary
181 Output a condensed summary of extended header information such as
182 file creations or deletions ("new" or "gone", optionally "+l" if
183 it’s a symlink) and mode changes ("+x" or "-x" for adding or
184 removing executable bit respectively) in diffstat. The information
185 is put between the filename part and the graph part. Implies
186 --stat.
187 --numstat
189 Similar to --stat, but shows number of added and deleted lines in
190 decimal notation and pathname without abbreviation, to make it more
191 machine friendly. For binary files, outputs two - instead of saying
192 0 0.
193 --shortstat
195 Output only the last line of the --stat format containing total
196 number of modified files, as well as number of added and deleted
197 lines.
198 -X[<param1,param2,...>], --dirstat[=<param1,param2,...>]
200 Output the distribution of relative amount of changes for each
201 sub-directory. The behavior of --dirstat can be customized by
202 passing it a comma separated list of parameters. The defaults are
203 controlled by the diff.dirstat configuration variable (see git-
204 config(1)). The following parameters are available:
205 changes
207 Compute the dirstat numbers by counting the lines that have
208 been removed from the source, or added to the destination. This
209 ignores the amount of pure code movements within a file. In
210 other words, rearranging lines in a file is not counted as much
211 as other changes. This is the default behavior when no
212 parameter is given.
213 lines
215 Compute the dirstat numbers by doing the regular line-based
216 diff analysis, and summing the removed/added line counts. (For
217 binary files, count 64-byte chunks instead, since binary files
218 have no natural concept of lines). This is a more expensive
219 --dirstat behavior than the changes behavior, but it does count
220 rearranged lines within a file as much as other changes. The
221 resulting output is consistent with what you get from the other
222 --*stat options.
223 files
225 Compute the dirstat numbers by counting the number of files
226 changed. Each changed file counts equally in the dirstat
227 analysis. This is the computationally cheapest --dirstat
228 behavior, since it does not have to look at the file contents
229 at all.
230 cumulative
232 Count changes in a child directory for the parent directory as
233 well. Note that when using cumulative, the sum of the
234 percentages reported may exceed 100%. The default
235 (non-cumulative) behavior can be specified with the
236 noncumulative parameter.
237 <limit>
239 An integer parameter specifies a cut-off percent (3% by
240 default). Directories contributing less than this percentage of
241 the changes are not shown in the output.
242 Example: The following will count changed files, while ignoring
244 directories with less than 10% of the total amount of changed
245 files, and accumulating child directory counts in the parent
246 directories: --dirstat=files,10,cumulative.
247 --cumulative
249 Synonym for --dirstat=cumulative
250 --dirstat-by-file[=<param1,param2>...]
252 Synonym for --dirstat=files,param1,param2...
253 --summary
255 Output a condensed summary of extended header information such as
256 creations, renames and mode changes.
257 --no-renames
259 Turn off rename detection, even when the configuration file gives
260 the default to do so.
261 --[no-]rename-empty
263 Whether to use empty blobs as rename source.
264 --full-index
266 Instead of the first handful of characters, show the full pre- and
267 post-image blob object names on the "index" line when generating
268 patch format output.
269 --binary
271 In addition to --full-index, output a binary diff that can be
272 applied with git-apply.
273 --abbrev[=<n>]
275 Instead of showing the full 40-byte hexadecimal object name in
276 diff-raw format output and diff-tree header lines, show the
277 shortest prefix that is at least <n> hexdigits long that uniquely
278 refers the object. In diff-patch output format, --full-index takes
279 higher precedence, i.e. if --full-index is specified, full blob
280 names will be shown regardless of --abbrev. Non default number of
281 digits can be specified with --abbrev=<n>.
282 -B[<n>][/<m>], --break-rewrites[=[<n>][/<m>]]
284 Break complete rewrite changes into pairs of delete and create.
285 This serves two purposes:
286 It affects the way a change that amounts to a total rewrite of a
288 file not as a series of deletion and insertion mixed together with
289 a very few lines that happen to match textually as the context, but
290 as a single deletion of everything old followed by a single
291 insertion of everything new, and the number m controls this aspect
292 of the -B option (defaults to 60%). -B/70% specifies that less
293 than 30% of the original should remain in the result for Git to
294 consider it a total rewrite (i.e. otherwise the resulting patch
295 will be a series of deletion and insertion mixed together with
296 context lines).
297 When used with -M, a totally-rewritten file is also considered as
299 the source of a rename (usually -M only considers a file that
300 disappeared as the source of a rename), and the number n controls
301 this aspect of the -B option (defaults to 50%). -B20% specifies
302 that a change with addition and deletion compared to 20% or more of
303 the file’s size are eligible for being picked up as a possible
304 source of a rename to another file.
305 -M[<n>], --find-renames[=<n>]
307 Detect renames. If n is specified, it is a threshold on the
308 similarity index (i.e. amount of addition/deletions compared to the
309 file’s size). For example, -M90% means Git should consider a
310 delete/add pair to be a rename if more than 90% of the file hasn’t
311 changed. Without a % sign, the number is to be read as a fraction,
312 with a decimal point before it. I.e., -M5 becomes 0.5, and is thus
313 the same as -M50%. Similarly, -M05 is the same as -M5%. To limit
314 detection to exact renames, use -M100%. The default similarity
315 index is 50%.
316 -C[<n>], --find-copies[=<n>]
318 Detect copies as well as renames. See also --find-copies-harder. If
319 n is specified, it has the same meaning as for -M<n>.
320 --find-copies-harder
322 For performance reasons, by default, -C option finds copies only if
323 the original file of the copy was modified in the same changeset.
324 This flag makes the command inspect unmodified files as candidates
325 for the source of copy. This is a very expensive operation for
326 large projects, so use it with caution. Giving more than one -C
327 option has the same effect.
328 -D, --irreversible-delete
330 Omit the preimage for deletes, i.e. print only the header but not
331 the diff between the preimage and /dev/null. The resulting patch is
332 not meant to be applied with patch or git apply; this is solely for
333 people who want to just concentrate on reviewing the text after the
334 change. In addition, the output obviously lacks enough information
335 to apply such a patch in reverse, even manually, hence the name of
336 the option.
337 When used together with -B, omit also the preimage in the deletion
339 part of a delete/create pair.
340 -l<num>
342 The -M and -C options involve some preliminary steps that can
343 detect subsets of renames/copies cheaply, followed by an exhaustive
344 fallback portion that compares all remaining unpaired destinations
345 to all relevant sources. (For renames, only remaining unpaired
346 sources are relevant; for copies, all original sources are
347 relevant.) For N sources and destinations, this exhaustive check is
348 O(N^2). This option prevents the exhaustive portion of rename/copy
349 detection from running if the number of source/destination files
350 involved exceeds the specified number. Defaults to
351 diff.renameLimit. Note that a value of 0 is treated as unlimited.
352 -O<orderfile>
354 Control the order in which files appear in the output. This
355 overrides the diff.orderFile configuration variable (see git-
356 config(1)). To cancel diff.orderFile, use -O/dev/null.
357 The output order is determined by the order of glob patterns in
359 <orderfile>. All files with pathnames that match the first pattern
360 are output first, all files with pathnames that match the second
361 pattern (but not the first) are output next, and so on. All files
362 with pathnames that do not match any pattern are output last, as if
363 there was an implicit match-all pattern at the end of the file. If
364 multiple pathnames have the same rank (they match the same pattern
365 but no earlier patterns), their output order relative to each other
366 is the normal order.
367 <orderfile> is parsed as follows:
369 • Blank lines are ignored, so they can be used as separators for
371 readability.
372 • Lines starting with a hash ("#") are ignored, so they can be
374 used for comments. Add a backslash ("\") to the beginning of
375 the pattern if it starts with a hash.
376 • Each other line contains a single pattern.
378 Patterns have the same syntax and semantics as patterns used for
380 fnmatch(3) without the FNM_PATHNAME flag, except a pathname also
381 matches a pattern if removing any number of the final pathname
382 components matches the pattern. For example, the pattern "foo*bar"
383 matches "fooasdfbar" and "foo/bar/baz/asdf" but not "foobarx".
384 --skip-to=<file>, --rotate-to=<file>
386 Discard the files before the named <file> from the output (i.e.
387 skip to), or move them to the end of the output (i.e. rotate to).
388 These were invented primarily for use of the git difftool command,
389 and may not be very useful otherwise.
390 --relative[=<path>], --no-relative
392 When run from a subdirectory of the project, it can be told to
393 exclude changes outside the directory and show pathnames relative
394 to it with this option. When you are not in a subdirectory (e.g. in
395 a bare repository), you can name which subdirectory to make the
396 output relative to by giving a <path> as an argument.
397 --no-relative can be used to countermand both diff.relative config
398 option and previous --relative.
399 -a, --text
401 Treat all files as text.
402 --ignore-cr-at-eol
404 Ignore carriage-return at the end of line when doing a comparison.
405 --ignore-space-at-eol
407 Ignore changes in whitespace at EOL.
408 -b, --ignore-space-change
410 Ignore changes in amount of whitespace. This ignores whitespace at
411 line end, and considers all other sequences of one or more
412 whitespace characters to be equivalent.
413 -w, --ignore-all-space
415 Ignore whitespace when comparing lines. This ignores differences
416 even if one line has whitespace where the other line has none.
417 --ignore-blank-lines
419 Ignore changes whose lines are all blank.
420 -I<regex>, --ignore-matching-lines=<regex>
422 Ignore changes whose all lines match <regex>. This option may be
423 specified more than once.
424 --inter-hunk-context=<lines>
426 Show the context between diff hunks, up to the specified number of
427 lines, thereby fusing hunks that are close to each other. Defaults
428 to diff.interHunkContext or 0 if the config option is unset.
429 -W, --function-context
431 Show whole function as context lines for each change. The function
432 names are determined in the same way as git diff works out patch
433 hunk headers (see Defining a custom hunk-header in
434 gitattributes(5)).
435 --ext-diff
437 Allow an external diff helper to be executed. If you set an
438 external diff driver with gitattributes(5), you need to use this
439 option with git-log(1) and friends.
440 --no-ext-diff
442 Disallow external diff drivers.
443 --textconv, --no-textconv
445 Allow (or disallow) external text conversion filters to be run when
446 comparing binary files. See gitattributes(5) for details. Because
447 textconv filters are typically a one-way conversion, the resulting
448 diff is suitable for human consumption, but cannot be applied. For
449 this reason, textconv filters are enabled by default only for git-
450 diff(1) and git-log(1), but not for git-format-patch(1) or diff
451 plumbing commands.
452 --ignore-submodules[=<when>]
454 Ignore changes to submodules in the diff generation. <when> can be
455 either "none", "untracked", "dirty" or "all", which is the default.
456 Using "none" will consider the submodule modified when it either
457 contains untracked or modified files or its HEAD differs from the
458 commit recorded in the superproject and can be used to override any
459 settings of the ignore option in git-config(1) or gitmodules(5).
460 When "untracked" is used submodules are not considered dirty when
461 they only contain untracked content (but they are still scanned for
462 modified content). Using "dirty" ignores all changes to the work
463 tree of submodules, only changes to the commits stored in the
464 superproject are shown (this was the behavior until 1.7.0). Using
465 "all" hides all changes to submodules.
466 --src-prefix=<prefix>
468 Show the given source prefix instead of "a/".
469 --dst-prefix=<prefix>
471 Show the given destination prefix instead of "b/".
472 --no-prefix
474 Do not show any source or destination prefix.
475 --line-prefix=<prefix>
477 Prepend an additional prefix to every line of output.
478 --ita-invisible-in-index
480 By default entries added by "git add -N" appear as an existing
481 empty file in "git diff" and a new file in "git diff --cached".
482 This option makes the entry appear as a new file in "git diff" and
483 non-existent in "git diff --cached". This option could be reverted
484 with --ita-visible-in-index. Both options are experimental and
485 could be removed in future.
486 For more detailed explanation on these common options, see also
488 gitdiffcore(7).
489 -<n>
491 Prepare patches from the topmost <n> commits.
492 -o <dir>, --output-directory <dir>
494 Use <dir> to store the resulting files, instead of the current
495 working directory.
496 -n, --numbered
498 Name output in [PATCH n/m] format, even with a single patch.
499 -N, --no-numbered
501 Name output in [PATCH] format.
502 --start-number <n>
504 Start numbering the patches at <n> instead of 1.
505 --numbered-files
507 Output file names will be a simple number sequence without the
508 default first line of the commit appended.
509 -k, --keep-subject
511 Do not strip/add [PATCH] from the first line of the commit log
512 message.
513 -s, --signoff
515 Add a Signed-off-by trailer to the commit message, using the
516 committer identity of yourself. See the signoff option in git-
517 commit(1) for more information.
518 --stdout
520 Print all commits to the standard output in mbox format, instead of
521 creating a file for each one.
522 --attach[=<boundary>]
524 Create multipart/mixed attachment, the first part of which is the
525 commit message and the patch itself in the second part, with
526 Content-Disposition: attachment.
527 --no-attach
529 Disable the creation of an attachment, overriding the configuration
530 setting.
531 --inline[=<boundary>]
533 Create multipart/mixed attachment, the first part of which is the
534 commit message and the patch itself in the second part, with
535 Content-Disposition: inline.
536 --thread[=<style>], --no-thread
538 Controls addition of In-Reply-To and References headers to make the
539 second and subsequent mails appear as replies to the first. Also
540 controls generation of the Message-Id header to reference.
541 The optional <style> argument can be either shallow or deep.
543 shallow threading makes every mail a reply to the head of the
544 series, where the head is chosen from the cover letter, the
545 --in-reply-to, and the first patch mail, in this order. deep
546 threading makes every mail a reply to the previous one.
547 The default is --no-thread, unless the format.thread configuration
549 is set. If --thread is specified without a style, it defaults to
550 the style specified by format.thread if any, or else shallow.
551 Beware that the default for git send-email is to thread emails
553 itself. If you want git format-patch to take care of threading, you
554 will want to ensure that threading is disabled for git send-email.
555 --in-reply-to=<message id>
557 Make the first mail (or all the mails with --no-thread) appear as a
558 reply to the given <message id>, which avoids breaking threads to
559 provide a new patch series.
560 --ignore-if-in-upstream
562 Do not include a patch that matches a commit in <until>..<since>.
563 This will examine all patches reachable from <since> but not from
564 <until> and compare them with the patches being generated, and any
565 patch that matches is ignored.
566 --cover-from-description=<mode>
568 Controls which parts of the cover letter will be automatically
569 populated using the branch’s description.
570 If <mode> is message or default, the cover letter subject will be
572 populated with placeholder text. The body of the cover letter will
573 be populated with the branch’s description. This is the default
574 mode when no configuration nor command line option is specified.
575 If <mode> is subject, the first paragraph of the branch description
577 will populate the cover letter subject. The remainder of the
578 description will populate the body of the cover letter.
579 If <mode> is auto, if the first paragraph of the branch description
581 is greater than 100 bytes, then the mode will be message, otherwise
582 subject will be used.
583 If <mode> is none, both the cover letter subject and body will be
585 populated with placeholder text.
586 --subject-prefix=<subject prefix>
588 Instead of the standard [PATCH] prefix in the subject line, instead
589 use [<subject prefix>]. This allows for useful naming of a patch
590 series, and can be combined with the --numbered option.
591 --filename-max-length=<n>
593 Instead of the standard 64 bytes, chomp the generated output
594 filenames at around <n> bytes (too short a value will be silently
595 raised to a reasonable length). Defaults to the value of the
596 format.filenameMaxLength configuration variable, or 64 if
597 unconfigured.
598 --rfc
600 Alias for --subject-prefix="RFC PATCH". RFC means "Request For
601 Comments"; use this when sending an experimental patch for
602 discussion rather than application.
603 -v <n>, --reroll-count=<n>
605 Mark the series as the <n>-th iteration of the topic. The output
606 filenames have v<n> prepended to them, and the subject prefix
607 ("PATCH" by default, but configurable via the --subject-prefix
608 option) has ` v<n>` appended to it. E.g. --reroll-count=4 may
609 produce v4-0001-add-makefile.patch file that has "Subject: [PATCH
610 v4 1/20] Add makefile" in it. <n> does not have to be an integer
611 (e.g. "--reroll-count=4.4", or "--reroll-count=4rev2" are allowed),
612 but the downside of using such a reroll-count is that the
613 range-diff/interdiff with the previous version does not state
614 exactly which version the new interation is compared against.
615 --to=<email>
617 Add a To: header to the email headers. This is in addition to any
618 configured headers, and may be used multiple times. The negated
619 form --no-to discards all To: headers added so far (from config or
620 command line).
621 --cc=<email>
623 Add a Cc: header to the email headers. This is in addition to any
624 configured headers, and may be used multiple times. The negated
625 form --no-cc discards all Cc: headers added so far (from config or
626 command line).
627 --from, --from=<ident>
629 Use ident in the From: header of each commit email. If the author
630 ident of the commit is not textually identical to the provided
631 ident, place a From: header in the body of the message with the
632 original author. If no ident is given, use the committer ident.
633 Note that this option is only useful if you are actually sending
635 the emails and want to identify yourself as the sender, but retain
636 the original author (and git am will correctly pick up the in-body
637 header). Note also that git send-email already handles this
638 transformation for you, and this option should not be used if you
639 are feeding the result to git send-email.
640 --add-header=<header>
642 Add an arbitrary header to the email headers. This is in addition
643 to any configured headers, and may be used multiple times. For
644 example, --add-header="Organization: git-foo". The negated form
645 --no-add-header discards all (To:, Cc:, and custom) headers added
646 so far from config or command line.
647 --[no-]cover-letter
649 In addition to the patches, generate a cover letter file containing
650 the branch description, shortlog and the overall diffstat. You can
651 fill in a description in the file before sending it out.
652 --encode-email-headers, --no-encode-email-headers
654 Encode email headers that have non-ASCII characters with
655 "Q-encoding" (described in RFC 2047), instead of outputting the
656 headers verbatim. Defaults to the value of the
657 format.encodeEmailHeaders configuration variable.
658 --interdiff=<previous>
660 As a reviewer aid, insert an interdiff into the cover letter, or as
661 commentary of the lone patch of a 1-patch series, showing the
662 differences between the previous version of the patch series and
663 the series currently being formatted. previous is a single
664 revision naming the tip of the previous series which shares a
665 common base with the series being formatted (for example git
666 format-patch --cover-letter --interdiff=feature/v1 -3 feature/v2).
667 --range-diff=<previous>
669 As a reviewer aid, insert a range-diff (see git-range-diff(1)) into
670 the cover letter, or as commentary of the lone patch of a 1-patch
671 series, showing the differences between the previous version of the
672 patch series and the series currently being formatted. previous
673 can be a single revision naming the tip of the previous series if
674 it shares a common base with the series being formatted (for
675 example git format-patch --cover-letter --range-diff=feature/v1 -3
676 feature/v2), or a revision range if the two versions of the series
677 are disjoint (for example git format-patch --cover-letter
678 --range-diff=feature/v1~3..feature/v1 -3 feature/v2).
679 Note that diff options passed to the command affect how the primary
681 product of format-patch is generated, and they are not passed to
682 the underlying range-diff machinery used to generate the
683 cover-letter material (this may change in the future).
684 --creation-factor=<percent>
686 Used with --range-diff, tweak the heuristic which matches up
687 commits between the previous and current series of patches by
688 adjusting the creation/deletion cost fudge factor. See git-range-
689 diff(1)) for details.
690 --notes[=<ref>], --no-notes
692 Append the notes (see git-notes(1)) for the commit after the
693 three-dash line.
694 The expected use case of this is to write supporting explanation
696 for the commit that does not belong to the commit log message
697 proper, and include it with the patch submission. While one can
698 simply write these explanations after format-patch has run but
699 before sending, keeping them as Git notes allows them to be
700 maintained between versions of the patch series (but see the
701 discussion of the notes.rewrite configuration options in git-
702 notes(1) to use this workflow).
703 The default is --no-notes, unless the format.notes configuration is
705 set.
706 --[no-]signature=<signature>
708 Add a signature to each message produced. Per RFC 3676 the
709 signature is separated from the body by a line with '-- ' on it. If
710 the signature option is omitted the signature defaults to the Git
711 version number.
712 --signature-file=<file>
714 Works just like --signature except the signature is read from a
715 file.
716 --suffix=.<sfx>
718 Instead of using .patch as the suffix for generated filenames, use
719 specified suffix. A common alternative is --suffix=.txt. Leaving
720 this empty will remove the .patch suffix.
721 Note that the leading character does not have to be a dot; for
723 example, you can use --suffix=-patch to get
724 0001-description-of-my-change-patch.
725 -q, --quiet
727 Do not print the names of the generated files to standard output.
728 --no-binary
730 Do not output contents of changes in binary files, instead display
731 a notice that those files changed. Patches generated using this
732 option cannot be applied properly, but they are still useful for
733 code review.
734 --zero-commit
736 Output an all-zero hash in each patch’s From header instead of the
737 hash of the commit.
738 --[no-]base[=<commit>]
740 Record the base tree information to identify the state the patch
741 series applies to. See the BASE TREE INFORMATION section below for
742 details. If <commit> is "auto", a base commit is automatically
743 chosen. The --no-base option overrides a format.useAutoBase
744 configuration.
745 --root
747 Treat the revision argument as a <revision range>, even if it is
748 just a single commit (that would normally be treated as a <since>).
749 Note that root commits included in the specified range are always
750 formatted as creation patches, independently of this flag.
751 --progress
753 Show progress reports on stderr as patches are generated.
754
756 You can specify extra mail header lines to be added to each message,
757 defaults for the subject prefix and file suffix, number patches when
758 outputting more than one patch, add "To:" or "Cc:" headers, configure
759 attachments, change the patch output directory, and sign off patches
760 with configuration variables.
761 [format]
763 headers = "Organization: git-foo\n"
764 subjectPrefix = CHANGE
765 suffix = .txt
766 numbered = auto
767 to = <email>
768 cc = <email>
769 attach [ = mime-boundary-string ]
770 signOff = true
771 outputDirectory = <directory>
772 coverLetter = auto
773 coverFromDescription = auto
774
776 The patch produced by git format-patch is in UNIX mailbox format, with
777 a fixed "magic" time stamp to indicate that the file is output from
778 format-patch rather than a real mailbox, like so:
779 From 8f72bad1baf19a53459661343e21d6491c3908d3 Mon Sep 17 00:00:00 2001
781 From: Tony Luck <tony.luck@intel.com>
782 Date: Tue, 13 Jul 2010 11:42:54 -0700
783 Subject: [PATCH] =?UTF-8?q?[IA64]=20Put=20ia64=20config=20files=20on=20the=20?=
784 =?UTF-8?q?Uwe=20Kleine-K=C3=B6nig=20diet?=
785 MIME-Version: 1.0
786 Content-Type: text/plain; charset=UTF-8
787 Content-Transfer-Encoding: 8bit
788 arch/arm config files were slimmed down using a python script
790 (See commit c2330e286f68f1c408b4aa6515ba49d57f05beae comment)
791 Do the same for ia64 so we can have sleek & trim looking
793 ...
794 Typically it will be placed in a MUA’s drafts folder, edited to add
796 timely commentary that should not go in the changelog after the three
797 dashes, and then sent as a message whose body, in our example, starts
798 with "arch/arm config files were...". On the receiving end, readers can
799 save interesting patches in a UNIX mailbox and apply them with git-
800 am(1).
801 When a patch is part of an ongoing discussion, the patch generated by
803 git format-patch can be tweaked to take advantage of the git am
804 --scissors feature. After your response to the discussion comes a line
805 that consists solely of "-- >8 --" (scissors and perforation), followed
806 by the patch with unnecessary header fields removed:
807 ...
809 > So we should do such-and-such.
810 Makes sense to me. How about this patch?
812 -- >8 --
814 Subject: [IA64] Put ia64 config files on the Uwe Kleine-König diet
815 arch/arm config files were slimmed down using a python script
817 ...
818 When sending a patch this way, most often you are sending your own
820 patch, so in addition to the "From $SHA1 $magic_timestamp" marker you
821 should omit From: and Date: lines from the patch file. The patch title
822 is likely to be different from the subject of the discussion the patch
823 is in response to, so it is likely that you would want to keep the
824 Subject: line, like the example above.
825 Checking for patch corruption
827 Many mailers if not set up properly will corrupt whitespace. Here are
828 two common types of corruption:
829 • Empty context lines that do not have any whitespace.
831 • Non-empty context lines that have one extra whitespace at the
833 beginning.
834 One way to test if your MUA is set up correctly is:
836 • Send the patch to yourself, exactly the way you would, except with
838 To: and Cc: lines that do not contain the list and maintainer
839 address.
840 • Save that patch to a file in UNIX mailbox format. Call it a.patch,
842 say.
843 • Apply it:
845 $ git fetch <project> master:test-apply
847 $ git switch test-apply
848 $ git restore --source=HEAD --staged --worktree :/
849 $ git am a.patch
850 If it does not apply correctly, there can be various reasons.
852 • The patch itself does not apply cleanly. That is bad but does not
854 have much to do with your MUA. You might want to rebase the patch
855 with git-rebase(1) before regenerating it in this case.
856 • The MUA corrupted your patch; "am" would complain that the patch
858 does not apply. Look in the .git/rebase-apply/ subdirectory and see
859 what patch file contains and check for the common corruption
860 patterns mentioned above.
861 • While at it, check the info and final-commit files as well. If what
863 is in final-commit is not exactly what you would want to see in the
864 commit log message, it is very likely that the receiver would end
865 up hand editing the log message when applying your patch. Things
866 like "Hi, this is my first patch.\n" in the patch e-mail should
867 come after the three-dash line that signals the end of the commit
868 message.
869
871 Here are some hints on how to successfully submit patches inline using
872 various mailers.
873 GMail
875 GMail does not have any way to turn off line wrapping in the web
876 interface, so it will mangle any emails that you send. You can however
877 use "git send-email" and send your patches through the GMail SMTP
878 server, or use any IMAP email client to connect to the google IMAP
879 server and forward the emails through that.
880 For hints on using git send-email to send your patches through the
882 GMail SMTP server, see the EXAMPLE section of git-send-email(1).
883 For hints on submission using the IMAP interface, see the EXAMPLE
885 section of git-imap-send(1).
886 Thunderbird
888 By default, Thunderbird will both wrap emails as well as flag them as
889 being format=flowed, both of which will make the resulting email
890 unusable by Git.
891 There are three different approaches: use an add-on to turn off line
893 wraps, configure Thunderbird to not mangle patches, or use an external
894 editor to keep Thunderbird from mangling the patches.
895 Approach #1 (add-on)
897 Install the Toggle Word Wrap add-on that is available from
898 https://addons.mozilla.org/thunderbird/addon/toggle-word-wrap/ It
899 adds a menu entry "Enable Word Wrap" in the composer’s "Options"
900 menu that you can tick off. Now you can compose the message as you
901 otherwise do (cut + paste, git format-patch | git imap-send, etc),
902 but you have to insert line breaks manually in any text that you
903 type.
904 Approach #2 (configuration)
906 Three steps:
907 1. Configure your mail server composition as plain text: Edit...
909 Account Settings...Composition & Addressing, uncheck "Compose
910 Messages in HTML".
911 2. Configure your general composition window to not wrap.
913 In Thunderbird 2: Edit..Preferences..Composition, wrap plain
915 text messages at 0
916 In Thunderbird 3: Edit..Preferences..Advanced..Config Editor.
918 Search for "mail.wrap_long_lines". Toggle it to make sure it is
919 set to false. Also, search for "mailnews.wraplength" and set
920 the value to 0.
921 3. Disable the use of format=flowed:
923 Edit..Preferences..Advanced..Config Editor. Search for
924 "mailnews.send_plaintext_flowed". Toggle it to make sure it is
925 set to false.
926 After that is done, you should be able to compose email as you
928 otherwise would (cut + paste, git format-patch | git imap-send,
929 etc), and the patches will not be mangled.
930 Approach #3 (external editor)
932 The following Thunderbird extensions are needed: AboutConfig from
933 http://aboutconfig.mozdev.org/ and External Editor from
934 http://globs.org/articles.php?lng=en&pg=8
935 1. Prepare the patch as a text file using your method of choice.
937 2. Before opening a compose window, use Edit→Account Settings to
939 uncheck the "Compose messages in HTML format" setting in the
940 "Composition & Addressing" panel of the account to be used to
941 send the patch.
942 3. In the main Thunderbird window, before you open the compose
944 window for the patch, use Tools→about:config to set the
945 following to the indicated values:
946 mailnews.send_plaintext_flowed => false
948 mailnews.wraplength => 0
949 4. Open a compose window and click the external editor icon.
951 5. In the external editor window, read in the patch file and exit
953 the editor normally.
954 Side note: it may be possible to do step 2 with about:config and
956 the following settings but no one’s tried yet.
957 mail.html_compose => false
959 mail.identity.default.compose_html => false
960 mail.identity.id?.compose_html => false
961 There is a script in contrib/thunderbird-patch-inline which can
963 help you include patches with Thunderbird in an easy way. To use
964 it, do the steps above and then use the script as the external
965 editor.
966 KMail
968 This should help you to submit patches inline using KMail.
969 1. Prepare the patch as a text file.
971 2. Click on New Mail.
973 3. Go under "Options" in the Composer window and be sure that "Word
975 wrap" is not set.
976 4. Use Message → Insert file... and insert the patch.
978 5. Back in the compose window: add whatever other text you wish to the
980 message, complete the addressing and subject fields, and press
981 send.
982
984 The base tree information block is used for maintainers or third party
985 testers to know the exact state the patch series applies to. It
986 consists of the base commit, which is a well-known commit that is part
987 of the stable part of the project history everybody else works off of,
988 and zero or more prerequisite patches, which are well-known patches in
989 flight that is not yet part of the base commit that need to be applied
990 on top of base commit in topological order before the patches can be
991 applied.
992 The base commit is shown as "base-commit: " followed by the 40-hex of
994 the commit object name. A prerequisite patch is shown as
995 "prerequisite-patch-id: " followed by the 40-hex patch id, which can be
996 obtained by passing the patch through the git patch-id --stable
997 command.
998 Imagine that on top of the public commit P, you applied well-known
1000 patches X, Y and Z from somebody else, and then built your three-patch
1001 series A, B, C, the history would be like:
1002 ---P---X---Y---Z---A---B---C
1004 With git format-patch --base=P -3 C (or variants thereof, e.g. with
1006 --cover-letter or using Z..C instead of -3 C to specify the range), the
1007 base tree information block is shown at the end of the first message
1008 the command outputs (either the first patch, or the cover letter), like
1009 this:
1010 base-commit: P
1012 prerequisite-patch-id: X
1013 prerequisite-patch-id: Y
1014 prerequisite-patch-id: Z
1015 For non-linear topology, such as
1017 ---P---X---A---M---C
1019 \ /
1020 Y---Z---B
1021 You can also use git format-patch --base=P -3 C to generate patches for
1023 A, B and C, and the identifiers for P, X, Y, Z are appended at the end
1024 of the first message.
1025 If set --base=auto in cmdline, it will track base commit automatically,
1027 the base commit will be the merge base of tip commit of the
1028 remote-tracking branch and revision-range specified in cmdline. For a
1029 local branch, you need to track a remote branch by git branch
1030 --set-upstream-to before using this option.
1031
1033 • Extract commits between revisions R1 and R2, and apply them on top
1034 of the current branch using git am to cherry-pick them:
1035 $ git format-patch -k --stdout R1..R2 | git am -3 -k
1037 • Extract all commits which are in the current branch but not in the
1039 origin branch:
1040 $ git format-patch origin
1042 For each commit a separate file is created in the current
1044 directory.
1045 • Extract all commits that lead to origin since the inception of the
1047 project:
1048 $ git format-patch --root origin
1050 • The same as the previous one:
1052 $ git format-patch -M -B origin
1054 Additionally, it detects and handles renames and complete rewrites
1056 intelligently to produce a renaming patch. A renaming patch reduces
1057 the amount of text output, and generally makes it easier to review.
1058 Note that non-Git "patch" programs won’t understand renaming
1059 patches, so use it only when you know the recipient uses Git to
1060 apply your patch.
1061 • Extract three topmost commits from the current branch and format
1063 them as e-mailable patches:
1064 $ git format-patch -3
1066
1068 Note that format-patch will omit merge commits from the output, even if
1069 they are part of the requested range. A simple "patch" does not include
1070 enough information for the receiving end to reproduce the same merge
1071 commit.
1072
1074 git-am(1), git-send-email(1)
1075
1077 Part of the git(1) suite
1078Git 2.33.1 2021-10-12 GIT-FORMAT-PATCH(1)