Do a cross-project merge of Paul Mackerras' gitk visualizer
[git/git.git] / Documentation / diffcore.txt
1 Tweaking diff output
2 ====================
3 June 2005
6 Introduction
7 ------------
9 The diff commands git-diff-cache, git-diff-files, and
10 git-diff-tree can be told to manipulate differences they find
11 in unconventional ways before showing diff(1) output. The
12 manipulation is collectively called "diffcore transformation".
13 This short note describes what they are and how to use them to
14 produce diff outputs that are easier to understand than the
15 conventional kind.
18 The chain of operation
19 ----------------------
21 The git-diff-* family works by first comparing two sets of
22 files:
24 - git-diff-cache compares contents of a "tree" object and the
25 working directory (when --cached flag is not used) or a
26 "tree" object and the index file (when --cached flag is
27 used);
29 - git-diff-files compares contents of the index file and the
30 working directory;
32 - git-diff-tree compares contents of two "tree" objects.
34 In all of these cases, the commands themselves compare
35 corresponding paths in the two sets of files. The result of
36 comparison is passed from these commands to what is internally
37 called "diffcore", in a format similar to what is output when
38 the -p option is not used. E.g.
40 in-place edit :100644 100644 bcd1234... 0123456... M file0
41 create :000000 100644 0000000... 1234567... N file4
42 delete :100644 000000 1234567... 0000000... D file5
43 unmerged :000000 000000 0000000... 0000000... U file6
45 The diffcore mechanism is fed a list of such comparison results
46 (each of which is called "filepair", although at this point each
47 of them talks about a single file), and transforms such a list
48 into another list. There are currently 6 such transformations:
50 - diffcore-pathspec
51 - diffcore-break
52 - diffcore-rename
53 - diffcore-merge-broken
54 - diffcore-pickaxe
55 - diffcore-order
57 These are applied in sequence. The set of filepairs git-diff-*
58 commands find are used as the input to diffcore-pathspec, and
59 the output from diffcore-pathspec is used as the input to the
60 next transformation. The final result is then passed to the
61 output routine and generates either diff-raw format (see Output
62 format sections of the manual for git-diff-* commands) or
63 diff-patch format.
66 diffcore-pathspec
67 -----------------
69 The first transformation in the chain is diffcore-pathspec, and
70 is controlled by giving the pathname parameters to the
71 git-diff-* commands on the command line. The pathspec is used
72 to limit the world diff operates in. It removes the filepairs
73 outside the specified set of pathnames.
75 Implementation note. For performance reasons, git-diff-tree
76 uses the pathname parameters on the command line to cull set of
77 filepairs it feeds the diffcore mechanism itself, and does not
78 use diffcore-pathspec, but the end result is the same.
81 diffcore-break
82 --------------
84 The second transformation in the chain is diffcore-break, and is
85 controlled by the -B option to the git-diff-* commands. This is
86 used to detect a filepair that represents "complete rewrite" and
87 break such filepair into two filepairs that represent delete and
88 create. E.g. If the input contained this filepair:
90 :100644 100644 bcd1234... 0123456... M file0
92 and if it detects that the file "file0" is completely rewritten,
93 it changes it to:
95 :100644 000000 bcd1234... 0000000... D file0
96 :000000 100644 0000000... 0123456... N file0
98 For the purpose of breaking a filepair, diffcore-break examines
99 the extent of changes between the contents of the files before
100 and after modification (i.e. the contents that have "bcd1234..."
101 and "0123456..." as their SHA1 content ID, in the above
102 example). The amount of deletion of original contents and
103 insertion of new material are added together, and if it exceeds
104 the "break score", the filepair is broken into two. The break
105 score defaults to 50% of the size of the smaller of the original
106 and the result (i.e. if the edit shrinks the file, the size of
107 the result is used; if the edit lengthens the file, the size of
108 the original is used), and can be customized by giving a number
109 after "-B" option (e.g. "-B75" to tell it to use 75%).
112 diffcore-rename
113 ---------------
115 This transformation is used to detect renames and copies, and is
116 controlled by the -M option (to detect renames) and the -C option
117 (to detect copies as well) to the git-diff-* commands. If the
118 input contained these filepairs:
120 :100644 000000 0123456... 0000000... D fileX
121 :000000 100644 0000000... 0123456... N file0
123 and the contents of the deleted file fileX is similar enough to
124 the contents of the created file file0, then rename detection
125 merges these filepairs and creates:
127 :100644 100644 0123456... 0123456... R100 fileX file0
129 When the "-C" option is used, the original contents of modified
130 files and contents of unchanged files are considered as
131 candidates of the source files in rename/copy operation, in
132 addition to the deleted files. If the input were like these
133 filepairs, that talk about a modified file fileY and a newly
134 created file file0:
136 :100644 100644 0123456... 1234567... M fileY
137 :000000 100644 0000000... 0123456... N file0
139 the original contents of fileY and the resulting contents of
140 file0 are compared, and if they are similar enough, they are
141 changed to:
143 :100644 100644 0123456... 1234567... M fileY
144 :100644 100644 0123456... 0123456... C100 fileY file0
146 In both rename and copy detection, the same "extent of changes"
147 algorithm used in diffcore-break is used to determine if two
148 files are "similar enough", and can be customized to use
149 similarity score different from the default 50% by giving a
150 number after "-M" or "-C" option (e.g. "-M8" to tell it to use
151 8/10 = 80%).
153 Note. When the "-C" option is used with --find-copies-harder
154 option, git-diff-* commands feed unmodified filepairs to
155 diffcore mechanism as well as modified ones. This lets the copy
156 detector consider unmodified files as copy source candidates at
157 the expense of making it slower. Without --find-copies-harder,
158 git-diff-* commands can detect copies only if the file that was
159 copied happened to have been modified in the same changeset.
162 diffcore-merge-broken
163 ---------------------
165 This transformation is used to merge filepairs broken by
166 diffcore-break, and were not transformed into rename/copy by
167 diffcore-rename, back into a single modification. This always
168 runs when diffcore-break is used.
170 For the purpose of merging broken filepairs back, it uses a
171 different "extent of changes" computation from the ones used by
172 diffcore-break and diffcore-rename. It counts only the deletion
173 from the original, and does not count insertion. If you removed
174 only 10 lines from a 100-line document, even if you added 910
175 new lines to make a new 1000-line document, you did not do a
176 complete rewrite. diffcore-break breaks such a case in order to
177 help diffcore-rename to consider such filepairs as candidate of
178 rename/copy detection, but if filepairs broken that way were not
179 matched with other filepairs to create rename/copy, then this
180 transformation merges them back into the original
181 "modification".
183 The "extent of changes" parameter can be tweaked from the
184 default 80% (that is, unless more than 80% of the original
185 material is deleted, the broken pairs are merged back into a
186 single modification) by giving a second number to -B option,
187 like these:
189 -B50/60 (give 50% "break score" to diffcore-break, use
190 60% for diffcore-merge-broken).
191 -B/60 (the same as above, since diffcore-break defautls to
192 50%).
194 Note that earlier implementation left a broken pair as a separate
195 creation and deletion patches. This was unnecessary hack and
196 the latest implementation always merges all the broken pairs
197 back into modifications, but the resulting patch output is
198 formatted differently to still let the reviewing easier for such
199 a complete rewrite by showing the entire contents of old version
200 prefixed with '-', followed by the entire contents of new
201 version prefixed with '+'.
204 diffcore-pickaxe
205 ----------------
207 This transformation is used to find filepairs that represent
208 changes that touch a specified string, and is controlled by the
209 -S option and the --pickaxe-all option to the git-diff-*
210 commands.
212 When diffcore-pickaxe is in use, it checks if there are
213 filepairs whose "original" side has the specified string and
214 whose "result" side does not. Such a filepair represents "the
215 string appeared in this changeset". It also checks for the
216 opposite case that loses the specified string.
218 When --pickaxe-all is not in effect, diffcore-pickaxe leaves
219 only such filepairs that touches the specified string in its
220 output. When --pickaxe-all is used, diffcore-pickaxe leaves all
221 filepairs intact if there is such a filepair, or makes the
222 output empty otherwise. The latter behaviour is designed to
223 make reviewing of the changes in the context of the whole
224 changeset easier.
227 diffcore-order
228 --------------
230 This is used to reorder the filepairs according to the user's
231 (or project's) taste, and is controlled by the -O option to the
232 git-diff-* commands.
234 This takes a text file each of whose line is a shell glob
235 pattern. Filepairs that match a glob pattern on an earlier line
236 in the file are output before ones that match a later line, and
237 filepairs that do not match any glob pattern are output last.
239 As an example, typical orderfile for the core GIT probably
240 should look like this:
243 Makefile
244 Documentation
245 *.h
246 *.c
247 t