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 ( <range1> <range2> | <rev1>...<rev2> | <base> <rev1> <rev2> )
13
15 This command shows the differences between two versions of a patch
16 series, or more generally, two commit ranges (ignoring merge commits).
17
18 To that end, it first finds pairs of commits from both commit ranges
19 that correspond with each other. Two commits are said to correspond
20 when the diff between their patches (i.e. the author information, the
21 commit message and the commit diff) is reasonably small compared to the
22 patches' size. See ``Algorithm`` below for details.
23
24 Finally, the list of matching commits is shown in the order of the
25 second commit range, with unmatched commits being inserted just after
26 all of their ancestors have been shown.
27
29 --no-dual-color
30 When the commit diffs differ, git range-diff recreates the original
31 diffs' coloring, and adds outer -/+ diff markers with the
32 background being red/green to make it easier to see e.g. when there
33 was a change in what exact lines were added.
34
35 Additionally, the commit diff lines that are only present in the
36 first commit range are shown "dimmed" (this can be overridden using
37 the color.diff.<slot> config setting where <slot> is one of
38 contextDimmed, oldDimmed and newDimmed), and the commit diff lines
39 that are only present in the second commit range are shown in bold
40 (which can be overridden using the config settings
41 color.diff.<slot> with <slot> being one of contextBold, oldBold or
42 newBold).
43
44 This is known to range-diff as "dual coloring". Use --no-dual-color
45 to revert to color all lines according to the outer diff markers
46 (and completely ignore the inner diff when it comes to color).
47
48 --creation-factor=<percent>
49 Set the creation/deletion cost fudge factor to <percent>. Defaults
50 to 60. Try a larger value if git range-diff erroneously considers a
51 large change a total rewrite (deletion of one commit and addition
52 of another), and a smaller one in the reverse case. See the
53 ``Algorithm`` section below for an explanation why this is needed.
54
55 --[no-]notes[=<ref>]
56 This flag is passed to the git log program (see git-log(1)) that
57 generates the patches.
58
59 <range1> <range2>
60 Compare the commits specified by the two ranges, where <range1> is
61 considered an older version of <range2>.
62
63 <rev1>...<rev2>
64 Equivalent to passing <rev2>..<rev1> and <rev1>..<rev2>.
65
66 <base> <rev1> <rev2>
67 Equivalent to passing <base>..<rev1> and <base>..<rev2>. Note that
68 <base> does not need to be the exact branch point of the branches.
69 Example: after rebasing a branch my-topic, git range-diff
70 my-topic@{u} my-topic@{1} my-topic would show the differences
71 introduced by the rebase.
72
73 git range-diff also accepts the regular diff options (see git-diff(1)),
74 most notably the --color=[<when>] and --no-color options. These options
75 are used when generating the "diff between patches", i.e. to compare
76 the author, commit message and diff of corresponding old/new commits.
77 There is currently no means to tweak most of the diff options passed to
78 git log when generating those patches.
79
81 The output of the range-diff command is subject to change. It is
82 intended to be human-readable porcelain output, not something that can
83 be used across versions of Git to get a textually stable range-diff (as
84 opposed to something like the --stable option to git-patch-id(1)).
85 There’s also no equivalent of git-apply(1) for range-diff, the output
86 is not intended to be machine-readable.
87
88 This is particularly true when passing in diff options. Currently some
89 options like --stat can, as an emergent effect, produce output that’s
90 quite useless in the context of range-diff. Future versions of
91 range-diff may learn to interpret such options in a manner specific to
92 range-diff (e.g. for --stat producing human-readable output which
93 summarizes how the diffstat changed).
94
96 This command uses the diff.color.* and pager.range-diff settings (the
97 latter is on by default). See git-config(1).
98
100 When a rebase required merge conflicts to be resolved, compare the
101 changes introduced by the rebase directly afterwards using:
102
103 $ git range-diff @{u} @{1} @
104
105 A typical output of git range-diff would look like this:
106
107 -: ------- > 1: 0ddba11 Prepare for the inevitable!
108 1: c0debee = 2: cab005e Add a helpful message at the start
109 2: f00dbal ! 3: decafe1 Describe a bug
110 @@ -1,3 +1,3 @@
111 Author: A U Thor <author@example.com>
112
113 -TODO: Describe a bug
114 +Describe a bug
115 @@ -324,5 +324,6
116 This is expected.
117
118 -+What is unexpected is that it will also crash.
119 ++Unexpectedly, it also crashes. This is a bug, and the jury is
120 ++still out there how to fix it best. See ticket #314 for details.
121
122 Contact
123 3: bedead < -: ------- TO-UNDO
124
125 In this example, there are 3 old and 3 new commits, where the developer
126 removed the 3rd, added a new one before the first two, and modified the
127 commit message of the 2nd commit as well its diff.
128
129 When the output goes to a terminal, it is color-coded by default, just
130 like regular git diff's output. In addition, the first line (adding a
131 commit) is green, the last line (deleting a commit) is red, the second
132 line (with a perfect match) is yellow like the commit header of git
133 show's output, and the third line colors the old commit red, the new
134 one green and the rest like git show's commit header.
135
136 A naive color-coded diff of diffs is actually a bit hard to read,
137 though, as it colors the entire lines red or green. The line that added
138 "What is unexpected" in the old commit, for example, is completely red,
139 even if the intent of the old commit was to add something.
140
141 To help with that, range uses the --dual-color mode by default. In this
142 mode, the diff of diffs will retain the original diff colors, and
143 prefix the lines with -/+ markers that have their background red or
144 green, to make it more obvious that they describe how the diff itself
145 changed.
146
148 The general idea is this: we generate a cost matrix between the commits
149 in both commit ranges, then solve the least-cost assignment.
150
151 The cost matrix is populated thusly: for each pair of commits, both
152 diffs are generated and the "diff of diffs" is generated, with 3
153 context lines, then the number of lines in that diff is used as cost.
154
155 To avoid false positives (e.g. when a patch has been removed, and an
156 unrelated patch has been added between two iterations of the same patch
157 series), the cost matrix is extended to allow for that, by adding
158 fixed-cost entries for wholesale deletes/adds.
159
160 Example: Let commits 1--2 be the first iteration of a patch series and
161 A--C the second iteration. Let’s assume that A is a cherry-pick of 2,
162 and C is a cherry-pick of 1 but with a small modification (say, a fixed
163 typo). Visualize the commits as a bipartite graph:
164
165 1 A
166
167 2 B
168
169 C
170
171 We are looking for a "best" explanation of the new series in terms of
172 the old one. We can represent an "explanation" as an edge in the graph:
173
174 1 A
175 /
176 2 --------' B
177
178 C
179
180 This explanation comes for "free" because there was no change.
181 Similarly C could be explained using 1, but that comes at some cost c>0
182 because of the modification:
183
184 1 ----. A
185 | /
186 2 ----+---' B
187 |
188 `----- C
189 c>0
190
191 In mathematical terms, what we are looking for is some sort of a
192 minimum cost bipartite matching; 1 is matched to C at some cost, etc.
193 The underlying graph is in fact a complete bipartite graph; the cost we
194 associate with every edge is the size of the diff between the two
195 commits' patches. To explain also new commits, we introduce dummy nodes
196 on both sides:
197
198 1 ----. A
199 | /
200 2 ----+---' B
201 |
202 o `----- C
203 c>0
204 o o
205
206 o o
207
208 The cost of an edge o--C is the size of C's diff, modified by a fudge
209 factor that should be smaller than 100%. The cost of an edge o--o is
210 free. The fudge factor is necessary because even if 1 and C have
211 nothing in common, they may still share a few empty lines and such,
212 possibly making the assignment 1--C, o--o slightly cheaper than 1--o,
213 o--C even if 1 and C have nothing in common. With the fudge factor we
214 require a much larger common part to consider patches as corresponding.
215
216 The overall time needed to compute this algorithm is the time needed to
217 compute n+m commit diffs and then n*m diffs of patches, plus the time
218 needed to compute the least-cost assignment between n and m diffs. Git
219 uses an implementation of the Jonker-Volgenant algorithm to solve the
220 assignment problem, which has cubic runtime complexity. The matching
221 found in this case will look like this:
222
223 1 ----. A
224 | /
225 2 ----+---' B
226 .--+-----'
227 o -' `----- C
228 c>0
229 o ---------- o
230
231 o ---------- o
232
234 git-log(1)
235
237 Part of the git(1) suite
238
239
240
241Git 2.30.2 2021-03-08 GIT-RANGE-DIFF(1)