1GIT-RANGE-DIFF(1) Git Manual GIT-RANGE-DIFF(1)
2
3
4
6 git-range-diff - Compare two commit ranges (e.g. two versions of a
7 branch)
8
10 git range-diff [--color=[<when>]] [--no-color] [<diff-options>]
11 [--no-dual-color] [--creation-factor=<factor>]
12 [--left-only | --right-only]
13 ( <range1> <range2> | <rev1>...<rev2> | <base> <rev1> <rev2> )
14
16 This command shows the differences between two versions of a patch
17 series, or more generally, two commit ranges (ignoring merge commits).
18
19 To that end, it first finds pairs of commits from both commit ranges
20 that correspond with each other. Two commits are said to correspond
21 when the diff between their patches (i.e. the author information, the
22 commit message and the commit diff) is reasonably small compared to the
23 patches' size. See ``Algorithm`` below for details.
24
25 Finally, the list of matching commits is shown in the order of the
26 second commit range, with unmatched commits being inserted just after
27 all of their ancestors have been shown.
28
29 There are three ways to specify the commit ranges:
30
31 • <range1> <range2>: Either commit range can be of the form
32 <base>..<rev>, <rev>^! or <rev>^-<n>. See SPECIFYING RANGES in
33 gitrevisions(7) for more details.
34
35 • <rev1>...<rev2>. This is equivalent to <rev2>..<rev1>
36 <rev1>..<rev2>.
37
38 • <base> <rev1> <rev2>: This is equivalent to <base>..<rev1>
39 <base>..<rev2>.
40
42 --no-dual-color
43 When the commit diffs differ, git range-diff recreates the original
44 diffs' coloring, and adds outer -/+ diff markers with the
45 background being red/green to make it easier to see e.g. when there
46 was a change in what exact lines were added.
47
48 Additionally, the commit diff lines that are only present in the
49 first commit range are shown "dimmed" (this can be overridden using
50 the color.diff.<slot> config setting where <slot> is one of
51 contextDimmed, oldDimmed and newDimmed), and the commit diff lines
52 that are only present in the second commit range are shown in bold
53 (which can be overridden using the config settings
54 color.diff.<slot> with <slot> being one of contextBold, oldBold or
55 newBold).
56
57 This is known to range-diff as "dual coloring". Use --no-dual-color
58 to revert to color all lines according to the outer diff markers
59 (and completely ignore the inner diff when it comes to color).
60
61 --creation-factor=<percent>
62 Set the creation/deletion cost fudge factor to <percent>. Defaults
63 to 60. Try a larger value if git range-diff erroneously considers a
64 large change a total rewrite (deletion of one commit and addition
65 of another), and a smaller one in the reverse case. See the
66 ``Algorithm`` section below for an explanation why this is needed.
67
68 --left-only
69 Suppress commits that are missing from the first specified range
70 (or the "left range" when using the <rev1>...<rev2> format).
71
72 --right-only
73 Suppress commits that are missing from the second specified range
74 (or the "right range" when using the <rev1>...<rev2> format).
75
76 --[no-]notes[=<ref>]
77 This flag is passed to the git log program (see git-log(1)) that
78 generates the patches.
79
80 <range1> <range2>
81 Compare the commits specified by the two ranges, where <range1> is
82 considered an older version of <range2>.
83
84 <rev1>...<rev2>
85 Equivalent to passing <rev2>..<rev1> and <rev1>..<rev2>.
86
87 <base> <rev1> <rev2>
88 Equivalent to passing <base>..<rev1> and <base>..<rev2>. Note that
89 <base> does not need to be the exact branch point of the branches.
90 Example: after rebasing a branch my-topic, git range-diff
91 my-topic@{u} my-topic@{1} my-topic would show the differences
92 introduced by the rebase.
93
94 git range-diff also accepts the regular diff options (see git-diff(1)),
95 most notably the --color=[<when>] and --no-color options. These options
96 are used when generating the "diff between patches", i.e. to compare
97 the author, commit message and diff of corresponding old/new commits.
98 There is currently no means to tweak most of the diff options passed to
99 git log when generating those patches.
100
102 The output of the range-diff command is subject to change. It is
103 intended to be human-readable porcelain output, not something that can
104 be used across versions of Git to get a textually stable range-diff (as
105 opposed to something like the --stable option to git-patch-id(1)).
106 There’s also no equivalent of git-apply(1) for range-diff, the output
107 is not intended to be machine-readable.
108
109 This is particularly true when passing in diff options. Currently some
110 options like --stat can, as an emergent effect, produce output that’s
111 quite useless in the context of range-diff. Future versions of
112 range-diff may learn to interpret such options in a manner specific to
113 range-diff (e.g. for --stat producing human-readable output which
114 summarizes how the diffstat changed).
115
117 This command uses the diff.color.* and pager.range-diff settings (the
118 latter is on by default). See git-config(1).
119
121 When a rebase required merge conflicts to be resolved, compare the
122 changes introduced by the rebase directly afterwards using:
123
124 $ git range-diff @{u} @{1} @
125
126 A typical output of git range-diff would look like this:
127
128 -: ------- > 1: 0ddba11 Prepare for the inevitable!
129 1: c0debee = 2: cab005e Add a helpful message at the start
130 2: f00dbal ! 3: decafe1 Describe a bug
131 @@ -1,3 +1,3 @@
132 Author: A U Thor <author@example.com>
133
134 -TODO: Describe a bug
135 +Describe a bug
136 @@ -324,5 +324,6
137 This is expected.
138
139 -+What is unexpected is that it will also crash.
140 ++Unexpectedly, it also crashes. This is a bug, and the jury is
141 ++still out there how to fix it best. See ticket #314 for details.
142
143 Contact
144 3: bedead < -: ------- TO-UNDO
145
146 In this example, there are 3 old and 3 new commits, where the developer
147 removed the 3rd, added a new one before the first two, and modified the
148 commit message of the 2nd commit as well its diff.
149
150 When the output goes to a terminal, it is color-coded by default, just
151 like regular git diff's output. In addition, the first line (adding a
152 commit) is green, the last line (deleting a commit) is red, the second
153 line (with a perfect match) is yellow like the commit header of git
154 show's output, and the third line colors the old commit red, the new
155 one green and the rest like git show's commit header.
156
157 A naive color-coded diff of diffs is actually a bit hard to read,
158 though, as it colors the entire lines red or green. The line that added
159 "What is unexpected" in the old commit, for example, is completely red,
160 even if the intent of the old commit was to add something.
161
162 To help with that, range uses the --dual-color mode by default. In this
163 mode, the diff of diffs will retain the original diff colors, and
164 prefix the lines with -/+ markers that have their background red or
165 green, to make it more obvious that they describe how the diff itself
166 changed.
167
169 The general idea is this: we generate a cost matrix between the commits
170 in both commit ranges, then solve the least-cost assignment.
171
172 The cost matrix is populated thusly: for each pair of commits, both
173 diffs are generated and the "diff of diffs" is generated, with 3
174 context lines, then the number of lines in that diff is used as cost.
175
176 To avoid false positives (e.g. when a patch has been removed, and an
177 unrelated patch has been added between two iterations of the same patch
178 series), the cost matrix is extended to allow for that, by adding
179 fixed-cost entries for wholesale deletes/adds.
180
181 Example: Let commits 1--2 be the first iteration of a patch series and
182 A--C the second iteration. Let’s assume that A is a cherry-pick of 2,
183 and C is a cherry-pick of 1 but with a small modification (say, a fixed
184 typo). Visualize the commits as a bipartite graph:
185
186 1 A
187
188 2 B
189
190 C
191
192 We are looking for a "best" explanation of the new series in terms of
193 the old one. We can represent an "explanation" as an edge in the graph:
194
195 1 A
196 /
197 2 --------' B
198
199 C
200
201 This explanation comes for "free" because there was no change.
202 Similarly C could be explained using 1, but that comes at some cost c>0
203 because of the modification:
204
205 1 ----. A
206 | /
207 2 ----+---' B
208 |
209 `----- C
210 c>0
211
212 In mathematical terms, what we are looking for is some sort of a
213 minimum cost bipartite matching; 1 is matched to C at some cost, etc.
214 The underlying graph is in fact a complete bipartite graph; the cost we
215 associate with every edge is the size of the diff between the two
216 commits' patches. To explain also new commits, we introduce dummy nodes
217 on both sides:
218
219 1 ----. A
220 | /
221 2 ----+---' B
222 |
223 o `----- C
224 c>0
225 o o
226
227 o o
228
229 The cost of an edge o--C is the size of C's diff, modified by a fudge
230 factor that should be smaller than 100%. The cost of an edge o--o is
231 free. The fudge factor is necessary because even if 1 and C have
232 nothing in common, they may still share a few empty lines and such,
233 possibly making the assignment 1--C, o--o slightly cheaper than 1--o,
234 o--C even if 1 and C have nothing in common. With the fudge factor we
235 require a much larger common part to consider patches as corresponding.
236
237 The overall time needed to compute this algorithm is the time needed to
238 compute n+m commit diffs and then n*m diffs of patches, plus the time
239 needed to compute the least-cost assignment between n and m diffs. Git
240 uses an implementation of the Jonker-Volgenant algorithm to solve the
241 assignment problem, which has cubic runtime complexity. The matching
242 found in this case will look like this:
243
244 1 ----. A
245 | /
246 2 ----+---' B
247 .--+-----'
248 o -' `----- C
249 c>0
250 o ---------- o
251
252 o ---------- o
253
255 git-log(1)
256
258 Part of the git(1) suite
259
260
261
262Git 2.33.1 2021-10-12 GIT-RANGE-DIFF(1)