Merge branch 'tr/rev-list-count'
[git/git.git] / Documentation / rev-list-options.txt
1 Commit Formatting
2 ~~~~~~~~~~~~~~~~~
3
4 ifdef::git-rev-list[]
5 Using these options, linkgit:git-rev-list[1] will act similar to the
6 more specialized family of commit log tools: linkgit:git-log[1],
7 linkgit:git-show[1], and linkgit:git-whatchanged[1]
8 endif::git-rev-list[]
9
10 include::pretty-options.txt[]
11
12 --relative-date::
13
14 Synonym for `--date=relative`.
15
16 --date={relative,local,default,iso,rfc,short,raw}::
17
18 Only takes effect for dates shown in human-readable format, such
19 as when using "--pretty". `log.date` config variable sets a default
20 value for log command's --date option.
21 +
22 `--date=relative` shows dates relative to the current time,
23 e.g. "2 hours ago".
24 +
25 `--date=local` shows timestamps in user's local timezone.
26 +
27 `--date=iso` (or `--date=iso8601`) shows timestamps in ISO 8601 format.
28 +
29 `--date=rfc` (or `--date=rfc2822`) shows timestamps in RFC 2822
30 format, often found in E-mail messages.
31 +
32 `--date=short` shows only date but not time, in `YYYY-MM-DD` format.
33 +
34 `--date=raw` shows the date in the internal raw git format `%s %z` format.
35 +
36 `--date=default` shows timestamps in the original timezone
37 (either committer's or author's).
38
39 ifdef::git-rev-list[]
40 --header::
41
42 Print the contents of the commit in raw-format; each record is
43 separated with a NUL character.
44 endif::git-rev-list[]
45
46 --parents::
47
48 Print the parents of the commit. Also enables parent
49 rewriting, see 'History Simplification' below.
50
51 --children::
52
53 Print the children of the commit. Also enables parent
54 rewriting, see 'History Simplification' below.
55
56 ifdef::git-rev-list[]
57 --timestamp::
58 Print the raw commit timestamp.
59 endif::git-rev-list[]
60
61 --left-right::
62
63 Mark which side of a symmetric diff a commit is reachable from.
64 Commits from the left side are prefixed with `<` and those from
65 the right with `>`. If combined with `--boundary`, those
66 commits are prefixed with `-`.
67 +
68 For example, if you have this topology:
69 +
70 -----------------------------------------------------------------------
71 y---b---b branch B
72 / \ /
73 / .
74 / / \
75 o---x---a---a branch A
76 -----------------------------------------------------------------------
77 +
78 you would get an output like this:
79 +
80 -----------------------------------------------------------------------
81 $ git rev-list --left-right --boundary --pretty=oneline A...B
82
83 >bbbbbbb... 3rd on b
84 >bbbbbbb... 2nd on b
85 <aaaaaaa... 3rd on a
86 <aaaaaaa... 2nd on a
87 -yyyyyyy... 1st on b
88 -xxxxxxx... 1st on a
89 -----------------------------------------------------------------------
90
91 --graph::
92
93 Draw a text-based graphical representation of the commit history
94 on the left hand side of the output. This may cause extra lines
95 to be printed in between commits, in order for the graph history
96 to be drawn properly.
97 +
98 This implies the '--topo-order' option by default, but the
99 '--date-order' option may also be specified.
100
101 ifdef::git-rev-list[]
102 --count::
103 Print a number stating how many commits would have been
104 listed, and suppress all other output. When used together
105 with '--left-right', instead print the counts for left and
106 right commits, separated by a tab.
107 endif::git-rev-list[]
108
109
110 ifndef::git-rev-list[]
111 Diff Formatting
112 ~~~~~~~~~~~~~~~
113
114 Below are listed options that control the formatting of diff output.
115 Some of them are specific to linkgit:git-rev-list[1], however other diff
116 options may be given. See linkgit:git-diff-files[1] for more options.
117
118 -c::
119
120 With this option, diff output for a merge commit
121 shows the differences from each of the parents to the merge result
122 simultaneously instead of showing pairwise diff between a parent
123 and the result one at a time. Furthermore, it lists only files
124 which were modified from all parents.
125
126 --cc::
127
128 This flag implies the '-c' options and further compresses the
129 patch output by omitting uninteresting hunks whose contents in
130 the parents have only two variants and the merge result picks
131 one of them without modification.
132
133 -m::
134
135 This flag makes the merge commits show the full diff like
136 regular commits; for each merge parent, a separate log entry
137 and diff is generated. An exception is that only diff against
138 the first parent is shown when '--first-parent' option is given;
139 in that case, the output represents the changes the merge
140 brought _into_ the then-current branch.
141
142 -r::
143
144 Show recursive diffs.
145
146 -t::
147
148 Show the tree objects in the diff output. This implies '-r'.
149 endif::git-rev-list[]
150
151 Commit Limiting
152 ~~~~~~~~~~~~~~~
153
154 Besides specifying a range of commits that should be listed using the
155 special notations explained in the description, additional commit
156 limiting may be applied.
157
158 --
159
160 -n 'number'::
161 --max-count=<number>::
162
163 Limit the number of commits output.
164
165 --skip=<number>::
166
167 Skip 'number' commits before starting to show the commit output.
168
169 --since=<date>::
170 --after=<date>::
171
172 Show commits more recent than a specific date.
173
174 --until=<date>::
175 --before=<date>::
176
177 Show commits older than a specific date.
178
179 ifdef::git-rev-list[]
180 --max-age=<timestamp>::
181 --min-age=<timestamp>::
182
183 Limit the commits output to specified time range.
184 endif::git-rev-list[]
185
186 --author=<pattern>::
187 --committer=<pattern>::
188
189 Limit the commits output to ones with author/committer
190 header lines that match the specified pattern (regular expression).
191
192 --grep=<pattern>::
193
194 Limit the commits output to ones with log message that
195 matches the specified pattern (regular expression).
196
197 --all-match::
198 Limit the commits output to ones that match all given --grep,
199 --author and --committer instead of ones that match at least one.
200
201 -i::
202 --regexp-ignore-case::
203
204 Match the regexp limiting patterns without regard to letters case.
205
206 -E::
207 --extended-regexp::
208
209 Consider the limiting patterns to be extended regular expressions
210 instead of the default basic regular expressions.
211
212 -F::
213 --fixed-strings::
214
215 Consider the limiting patterns to be fixed strings (don't interpret
216 pattern as a regular expression).
217
218 --remove-empty::
219
220 Stop when a given path disappears from the tree.
221
222 --merges::
223
224 Print only merge commits.
225
226 --no-merges::
227
228 Do not print commits with more than one parent.
229
230 --first-parent::
231 Follow only the first parent commit upon seeing a merge
232 commit. This option can give a better overview when
233 viewing the evolution of a particular topic branch,
234 because merges into a topic branch tend to be only about
235 adjusting to updated upstream from time to time, and
236 this option allows you to ignore the individual commits
237 brought in to your history by such a merge.
238
239 --not::
240
241 Reverses the meaning of the '{caret}' prefix (or lack thereof)
242 for all following revision specifiers, up to the next '--not'.
243
244 --all::
245
246 Pretend as if all the refs in `refs/` are listed on the
247 command line as '<commit>'.
248
249 --branches[=pattern]::
250
251 Pretend as if all the refs in `refs/heads` are listed
252 on the command line as '<commit>'. If `pattern` is given, limit
253 branches to ones matching given shell glob. If pattern lacks '?',
254 '*', or '[', '/*' at the end is implied.
255
256 --tags[=pattern]::
257
258 Pretend as if all the refs in `refs/tags` are listed
259 on the command line as '<commit>'. If `pattern` is given, limit
260 tags to ones matching given shell glob. If pattern lacks '?', '*',
261 or '[', '/*' at the end is implied.
262
263 --remotes[=pattern]::
264
265 Pretend as if all the refs in `refs/remotes` are listed
266 on the command line as '<commit>'. If `pattern`is given, limit
267 remote tracking branches to ones matching given shell glob.
268 If pattern lacks '?', '*', or '[', '/*' at the end is implied.
269
270 --glob=glob-pattern::
271 Pretend as if all the refs matching shell glob `glob-pattern`
272 are listed on the command line as '<commit>'. Leading 'refs/',
273 is automatically prepended if missing. If pattern lacks '?', '*',
274 or '[', '/*' at the end is implied.
275
276
277 ifndef::git-rev-list[]
278 --bisect::
279
280 Pretend as if the bad bisection ref `refs/bisect/bad`
281 was listed and as if it was followed by `--not` and the good
282 bisection refs `refs/bisect/good-*` on the command
283 line.
284 endif::git-rev-list[]
285
286 --stdin::
287
288 In addition to the '<commit>' listed on the command
289 line, read them from the standard input. If a '--' separator is
290 seen, stop reading commits and start reading paths to limit the
291 result.
292
293 ifdef::git-rev-list[]
294 --quiet::
295
296 Don't print anything to standard output. This form
297 is primarily meant to allow the caller to
298 test the exit status to see if a range of objects is fully
299 connected (or not). It is faster than redirecting stdout
300 to /dev/null as the output does not have to be formatted.
301 endif::git-rev-list[]
302
303 --cherry-pick::
304
305 Omit any commit that introduces the same change as
306 another commit on the "other side" when the set of
307 commits are limited with symmetric difference.
308 +
309 For example, if you have two branches, `A` and `B`, a usual way
310 to list all commits on only one side of them is with
311 `--left-right`, like the example above in the description of
312 that option. It however shows the commits that were cherry-picked
313 from the other branch (for example, "3rd on b" may be cherry-picked
314 from branch A). With this option, such pairs of commits are
315 excluded from the output.
316
317 -g::
318 --walk-reflogs::
319
320 Instead of walking the commit ancestry chain, walk
321 reflog entries from the most recent one to older ones.
322 When this option is used you cannot specify commits to
323 exclude (that is, '{caret}commit', 'commit1..commit2',
324 nor 'commit1...commit2' notations cannot be used).
325 +
326 With '\--pretty' format other than oneline (for obvious reasons),
327 this causes the output to have two extra lines of information
328 taken from the reflog. By default, 'commit@\{Nth}' notation is
329 used in the output. When the starting commit is specified as
330 'commit@\{now}', output also uses 'commit@\{timestamp}' notation
331 instead. Under '\--pretty=oneline', the commit message is
332 prefixed with this information on the same line.
333 This option cannot be combined with '\--reverse'.
334 See also linkgit:git-reflog[1].
335
336 --merge::
337
338 After a failed merge, show refs that touch files having a
339 conflict and don't exist on all heads to merge.
340
341 --boundary::
342
343 Output uninteresting commits at the boundary, which are usually
344 not shown.
345
346 --
347
348 History Simplification
349 ~~~~~~~~~~~~~~~~~~~~~~
350
351 Sometimes you are only interested in parts of the history, for example the
352 commits modifying a particular <path>. But there are two parts of
353 'History Simplification', one part is selecting the commits and the other
354 is how to do it, as there are various strategies to simplify the history.
355
356 The following options select the commits to be shown:
357
358 <paths>::
359
360 Commits modifying the given <paths> are selected.
361
362 --simplify-by-decoration::
363
364 Commits that are referred by some branch or tag are selected.
365
366 Note that extra commits can be shown to give a meaningful history.
367
368 The following options affect the way the simplification is performed:
369
370 Default mode::
371
372 Simplifies the history to the simplest history explaining the
373 final state of the tree. Simplest because it prunes some side
374 branches if the end result is the same (i.e. merging branches
375 with the same content)
376
377 --full-history::
378
379 As the default mode but does not prune some history.
380
381 --dense::
382
383 Only the selected commits are shown, plus some to have a
384 meaningful history.
385
386 --sparse::
387
388 All commits in the simplified history are shown.
389
390 --simplify-merges::
391
392 Additional option to '--full-history' to remove some needless
393 merges from the resulting history, as there are no selected
394 commits contributing to this merge.
395
396 --ancestry-path::
397
398 When given a range of commits to display (e.g. 'commit1..commit2'
399 or 'commit2 {caret}commit1'), only display commits that exist
400 directly on the ancestry chain between the 'commit1' and
401 'commit2', i.e. commits that are both descendants of 'commit1',
402 and ancestors of 'commit2'.
403
404 A more detailed explanation follows.
405
406 Suppose you specified `foo` as the <paths>. We shall call commits
407 that modify `foo` !TREESAME, and the rest TREESAME. (In a diff
408 filtered for `foo`, they look different and equal, respectively.)
409
410 In the following, we will always refer to the same example history to
411 illustrate the differences between simplification settings. We assume
412 that you are filtering for a file `foo` in this commit graph:
413 -----------------------------------------------------------------------
414 .-A---M---N---O---P
415 / / / / /
416 I B C D E
417 \ / / / /
418 `-------------'
419 -----------------------------------------------------------------------
420 The horizontal line of history A--P is taken to be the first parent of
421 each merge. The commits are:
422
423 * `I` is the initial commit, in which `foo` exists with contents
424 "asdf", and a file `quux` exists with contents "quux". Initial
425 commits are compared to an empty tree, so `I` is !TREESAME.
426
427 * In `A`, `foo` contains just "foo".
428
429 * `B` contains the same change as `A`. Its merge `M` is trivial and
430 hence TREESAME to all parents.
431
432 * `C` does not change `foo`, but its merge `N` changes it to "foobar",
433 so it is not TREESAME to any parent.
434
435 * `D` sets `foo` to "baz". Its merge `O` combines the strings from
436 `N` and `D` to "foobarbaz"; i.e., it is not TREESAME to any parent.
437
438 * `E` changes `quux` to "xyzzy", and its merge `P` combines the
439 strings to "quux xyzzy". Despite appearing interesting, `P` is
440 TREESAME to all parents.
441
442 'rev-list' walks backwards through history, including or excluding
443 commits based on whether '\--full-history' and/or parent rewriting
444 (via '\--parents' or '\--children') are used. The following settings
445 are available.
446
447 Default mode::
448
449 Commits are included if they are not TREESAME to any parent
450 (though this can be changed, see '\--sparse' below). If the
451 commit was a merge, and it was TREESAME to one parent, follow
452 only that parent. (Even if there are several TREESAME
453 parents, follow only one of them.) Otherwise, follow all
454 parents.
455 +
456 This results in:
457 +
458 -----------------------------------------------------------------------
459 .-A---N---O
460 / / /
461 I---------D
462 -----------------------------------------------------------------------
463 +
464 Note how the rule to only follow the TREESAME parent, if one is
465 available, removed `B` from consideration entirely. `C` was
466 considered via `N`, but is TREESAME. Root commits are compared to an
467 empty tree, so `I` is !TREESAME.
468 +
469 Parent/child relations are only visible with --parents, but that does
470 not affect the commits selected in default mode, so we have shown the
471 parent lines.
472
473 --full-history without parent rewriting::
474
475 This mode differs from the default in one point: always follow
476 all parents of a merge, even if it is TREESAME to one of them.
477 Even if more than one side of the merge has commits that are
478 included, this does not imply that the merge itself is! In
479 the example, we get
480 +
481 -----------------------------------------------------------------------
482 I A B N D O
483 -----------------------------------------------------------------------
484 +
485 `P` and `M` were excluded because they are TREESAME to a parent. `E`,
486 `C` and `B` were all walked, but only `B` was !TREESAME, so the others
487 do not appear.
488 +
489 Note that without parent rewriting, it is not really possible to talk
490 about the parent/child relationships between the commits, so we show
491 them disconnected.
492
493 --full-history with parent rewriting::
494
495 Ordinary commits are only included if they are !TREESAME
496 (though this can be changed, see '\--sparse' below).
497 +
498 Merges are always included. However, their parent list is rewritten:
499 Along each parent, prune away commits that are not included
500 themselves. This results in
501 +
502 -----------------------------------------------------------------------
503 .-A---M---N---O---P
504 / / / / /
505 I B / D /
506 \ / / / /
507 `-------------'
508 -----------------------------------------------------------------------
509 +
510 Compare to '\--full-history' without rewriting above. Note that `E`
511 was pruned away because it is TREESAME, but the parent list of P was
512 rewritten to contain `E`'s parent `I`. The same happened for `C` and
513 `N`. Note also that `P` was included despite being TREESAME.
514
515 In addition to the above settings, you can change whether TREESAME
516 affects inclusion:
517
518 --dense::
519
520 Commits that are walked are included if they are not TREESAME
521 to any parent.
522
523 --sparse::
524
525 All commits that are walked are included.
526 +
527 Note that without '\--full-history', this still simplifies merges: if
528 one of the parents is TREESAME, we follow only that one, so the other
529 sides of the merge are never walked.
530
531 --simplify-merges::
532
533 First, build a history graph in the same way that
534 '\--full-history' with parent rewriting does (see above).
535 +
536 Then simplify each commit `C` to its replacement `C'` in the final
537 history according to the following rules:
538 +
539 --
540 * Set `C'` to `C`.
541 +
542 * Replace each parent `P` of `C'` with its simplification `P'`. In
543 the process, drop parents that are ancestors of other parents, and
544 remove duplicates.
545 +
546 * If after this parent rewriting, `C'` is a root or merge commit (has
547 zero or >1 parents), a boundary commit, or !TREESAME, it remains.
548 Otherwise, it is replaced with its only parent.
549 --
550 +
551 The effect of this is best shown by way of comparing to
552 '\--full-history' with parent rewriting. The example turns into:
553 +
554 -----------------------------------------------------------------------
555 .-A---M---N---O
556 / / /
557 I B D
558 \ / /
559 `---------'
560 -----------------------------------------------------------------------
561 +
562 Note the major differences in `N` and `P` over '\--full-history':
563 +
564 --
565 * `N`'s parent list had `I` removed, because it is an ancestor of the
566 other parent `M`. Still, `N` remained because it is !TREESAME.
567 +
568 * `P`'s parent list similarly had `I` removed. `P` was then
569 removed completely, because it had one parent and is TREESAME.
570 --
571
572 Finally, there is a fifth simplification mode available:
573
574 --ancestry-path::
575
576 Limit the displayed commits to those directly on the ancestry
577 chain between the "from" and "to" commits in the given commit
578 range. I.e. only display commits that are ancestor of the "to"
579 commit, and descendants of the "from" commit.
580 +
581 As an example use case, consider the following commit history:
582 +
583 -----------------------------------------------------------------------
584 D---E-------F
585 / \ \
586 B---C---G---H---I---J
587 / \
588 A-------K---------------L--M
589 -----------------------------------------------------------------------
590 +
591 A regular 'D..M' computes the set of commits that are ancestors of `M`,
592 but excludes the ones that are ancestors of `D`. This is useful to see
593 what happened to the history leading to `M` since `D`, in the sense
594 that "what does `M` have that did not exist in `D`". The result in this
595 example would be all the commits, except `A` and `B` (and `D` itself,
596 of course).
597 +
598 When we want to find out what commits in `M` are contaminated with the
599 bug introduced by `D` and need fixing, however, we might want to view
600 only the subset of 'D..M' that are actually descendants of `D`, i.e.
601 excluding `C` and `K`. This is exactly what the '\--ancestry-path'
602 option does. Applied to the 'D..M' range, it results in:
603 +
604 -----------------------------------------------------------------------
605 E-------F
606 \ \
607 G---H---I---J
608 \
609 L--M
610 -----------------------------------------------------------------------
611
612 The '\--simplify-by-decoration' option allows you to view only the
613 big picture of the topology of the history, by omitting commits
614 that are not referenced by tags. Commits are marked as !TREESAME
615 (in other words, kept after history simplification rules described
616 above) if (1) they are referenced by tags, or (2) they change the
617 contents of the paths given on the command line. All other
618 commits are marked as TREESAME (subject to be simplified away).
619
620 ifdef::git-rev-list[]
621 Bisection Helpers
622 ~~~~~~~~~~~~~~~~~
623
624 --bisect::
625
626 Limit output to the one commit object which is roughly halfway between
627 included and excluded commits. Note that the bad bisection ref
628 `refs/bisect/bad` is added to the included commits (if it
629 exists) and the good bisection refs `refs/bisect/good-*` are
630 added to the excluded commits (if they exist). Thus, supposing there
631 are no refs in `refs/bisect/`, if
632
633 -----------------------------------------------------------------------
634 $ git rev-list --bisect foo ^bar ^baz
635 -----------------------------------------------------------------------
636
637 outputs 'midpoint', the output of the two commands
638
639 -----------------------------------------------------------------------
640 $ git rev-list foo ^midpoint
641 $ git rev-list midpoint ^bar ^baz
642 -----------------------------------------------------------------------
643
644 would be of roughly the same length. Finding the change which
645 introduces a regression is thus reduced to a binary search: repeatedly
646 generate and test new 'midpoint's until the commit chain is of length
647 one.
648
649 --bisect-vars::
650
651 This calculates the same as `--bisect`, except that refs in
652 `refs/bisect/` are not used, and except that this outputs
653 text ready to be eval'ed by the shell. These lines will assign the
654 name of the midpoint revision to the variable `bisect_rev`, and the
655 expected number of commits to be tested after `bisect_rev` is tested
656 to `bisect_nr`, the expected number of commits to be tested if
657 `bisect_rev` turns out to be good to `bisect_good`, the expected
658 number of commits to be tested if `bisect_rev` turns out to be bad to
659 `bisect_bad`, and the number of commits we are bisecting right now to
660 `bisect_all`.
661
662 --bisect-all::
663
664 This outputs all the commit objects between the included and excluded
665 commits, ordered by their distance to the included and excluded
666 commits. Refs in `refs/bisect/` are not used. The farthest
667 from them is displayed first. (This is the only one displayed by
668 `--bisect`.)
669 +
670 This is useful because it makes it easy to choose a good commit to
671 test when you want to avoid to test some of them for some reason (they
672 may not compile for example).
673 +
674 This option can be used along with `--bisect-vars`, in this case,
675 after all the sorted commit objects, there will be the same text as if
676 `--bisect-vars` had been used alone.
677 endif::git-rev-list[]
678
679
680 Commit Ordering
681 ~~~~~~~~~~~~~~~
682
683 By default, the commits are shown in reverse chronological order.
684
685 --topo-order::
686
687 This option makes them appear in topological order (i.e.
688 descendant commits are shown before their parents).
689
690 --date-order::
691
692 This option is similar to '--topo-order' in the sense that no
693 parent comes before all of its children, but otherwise things
694 are still ordered in the commit timestamp order.
695
696 --reverse::
697
698 Output the commits in reverse order.
699 Cannot be combined with '\--walk-reflogs'.
700
701 Object Traversal
702 ~~~~~~~~~~~~~~~~
703
704 These options are mostly targeted for packing of git repositories.
705
706 --objects::
707
708 Print the object IDs of any object referenced by the listed
709 commits. '--objects foo ^bar' thus means "send me
710 all object IDs which I need to download if I have the commit
711 object 'bar', but not 'foo'".
712
713 --objects-edge::
714
715 Similar to '--objects', but also print the IDs of excluded
716 commits prefixed with a "-" character. This is used by
717 linkgit:git-pack-objects[1] to build "thin" pack, which records
718 objects in deltified form based on objects contained in these
719 excluded commits to reduce network traffic.
720
721 --unpacked::
722
723 Only useful with '--objects'; print the object IDs that are not
724 in packs.
725
726 --no-walk::
727
728 Only show the given revs, but do not traverse their ancestors.
729
730 --do-walk::
731
732 Overrides a previous --no-walk.