1gitdiffcore(7) 2============== 3 4NAME 5---- 6gitdiffcore - Tweaking diff output (June 2005) 7 8SYNOPSIS 9-------- 10'git diff' * 11 12DESCRIPTION 13----------- 14 15The diff commands 'git-diff-index', 'git-diff-files', and 'git-diff-tree' 16can be told to manipulate differences they find in 17unconventional ways before showing 'diff' output. The manipulation 18is collectively called "diffcore transformation". This short note 19describes what they are and how to use them to produce 'diff' output 20that is easier to understand than the conventional kind. 21 22 23The chain of operation 24---------------------- 25 26The 'git-diff-{asterisk}' family works by first comparing two sets of 27files: 28 29 - 'git-diff-index' compares contents of a "tree" object and the 30 working directory (when '\--cached' flag is not used) or a 31 "tree" object and the index file (when '\--cached' flag is 32 used); 33 34 - 'git-diff-files' compares contents of the index file and the 35 working directory; 36 37 - 'git-diff-tree' compares contents of two "tree" objects; 38 39In all of these cases, the commands themselves compare 40corresponding paths in the two sets of files. The result of 41comparison is passed from these commands to what is internally 42called "diffcore", in a format similar to what is output when 43the -p option is not used. E.g. 44 45------------------------------------------------ 46in-place edit :100644 100644 bcd1234... 0123456... M file0 47create :000000 100644 0000000... 1234567... A file4 48delete :100644 000000 1234567... 0000000... D file5 49unmerged :000000 000000 0000000... 0000000... U file6 50------------------------------------------------ 51 52The diffcore mechanism is fed a list of such comparison results 53(each of which is called "filepair", although at this point each 54of them talks about a single file), and transforms such a list 55into another list. There are currently 6 such transformations: 56 57- diffcore-pathspec 58- diffcore-break 59- diffcore-rename 60- diffcore-merge-broken 61- diffcore-pickaxe 62- diffcore-order 63 64These are applied in sequence. The set of filepairs 'git-diff-{asterisk}' 65commands find are used as the input to diffcore-pathspec, and 66the output from diffcore-pathspec is used as the input to the 67next transformation. The final result is then passed to the 68output routine and generates either diff-raw format (see Output 69format sections of the manual for 'git-diff-{asterisk}' commands) or 70diff-patch format. 71 72 73diffcore-pathspec: For Ignoring Files Outside Our Consideration 74--------------------------------------------------------------- 75 76The first transformation in the chain is diffcore-pathspec, and 77is controlled by giving the pathname parameters to the 78'git-diff-{asterisk}' commands on the command line. The pathspec is used 79to limit the world diff operates in. It removes the filepairs 80outside the specified set of pathnames. E.g. If the input set 81of filepairs included: 82 83------------------------------------------------ 84:100644 100644 bcd1234... 0123456... M junkfile 85------------------------------------------------ 86 87but the command invocation was `git diff-files myfile`, then the 88junkfile entry would be removed from the list because only "myfile" 89is under consideration. 90 91Implementation note. For performance reasons, 'git-diff-tree' 92uses the pathname parameters on the command line to cull set of 93filepairs it feeds the diffcore mechanism itself, and does not 94use diffcore-pathspec, but the end result is the same. 95 96 97diffcore-break: For Splitting Up "Complete Rewrites" 98---------------------------------------------------- 99 100The second transformation in the chain is diffcore-break, and is 101controlled by the -B option to the 'git-diff-{asterisk}' commands. This is 102used to detect a filepair that represents "complete rewrite" and 103break such filepair into two filepairs that represent delete and 104create. E.g. If the input contained this filepair: 105 106------------------------------------------------ 107:100644 100644 bcd1234... 0123456... M file0 108------------------------------------------------ 109 110and if it detects that the file "file0" is completely rewritten, 111it changes it to: 112 113------------------------------------------------ 114:100644 000000 bcd1234... 0000000... D file0 115:000000 100644 0000000... 0123456... A file0 116------------------------------------------------ 117 118For the purpose of breaking a filepair, diffcore-break examines 119the extent of changes between the contents of the files before 120and after modification (i.e. the contents that have "bcd1234..." 121and "0123456..." as their SHA1 content ID, in the above 122example). The amount of deletion of original contents and 123insertion of new material are added together, and if it exceeds 124the "break score", the filepair is broken into two. The break 125score defaults to 50% of the size of the smaller of the original 126and the result (i.e. if the edit shrinks the file, the size of 127the result is used; if the edit lengthens the file, the size of 128the original is used), and can be customized by giving a number 129after "-B" option (e.g. "-B75" to tell it to use 75%). 130 131 132diffcore-rename: For Detection Renames and Copies 133------------------------------------------------- 134 135This transformation is used to detect renames and copies, and is 136controlled by the -M option (to detect renames) and the -C option 137(to detect copies as well) to the 'git-diff-{asterisk}' commands. If the 138input contained these filepairs: 139 140------------------------------------------------ 141:100644 000000 0123456... 0000000... D fileX 142:000000 100644 0000000... 0123456... A file0 143------------------------------------------------ 144 145and the contents of the deleted file fileX is similar enough to 146the contents of the created file file0, then rename detection 147merges these filepairs and creates: 148 149------------------------------------------------ 150:100644 100644 0123456... 0123456... R100 fileX file0 151------------------------------------------------ 152 153When the "-C" option is used, the original contents of modified files, 154and deleted files (and also unmodified files, if the 155"\--find-copies-harder" option is used) are considered as candidates 156of the source files in rename/copy operation. If the input were like 157these filepairs, that talk about a modified file fileY and a newly 158created file file0: 159 160------------------------------------------------ 161:100644 100644 0123456... 1234567... M fileY 162:000000 100644 0000000... bcd3456... A file0 163------------------------------------------------ 164 165the original contents of fileY and the resulting contents of 166file0 are compared, and if they are similar enough, they are 167changed to: 168 169------------------------------------------------ 170:100644 100644 0123456... 1234567... M fileY 171:100644 100644 0123456... bcd3456... C100 fileY file0 172------------------------------------------------ 173 174In both rename and copy detection, the same "extent of changes" 175algorithm used in diffcore-break is used to determine if two 176files are "similar enough", and can be customized to use 177a similarity score different from the default of 50% by giving a 178number after the "-M" or "-C" option (e.g. "-M8" to tell it to use 1798/10 = 80%). 180 181Note. When the "-C" option is used with `\--find-copies-harder` 182option, 'git-diff-{asterisk}' commands feed unmodified filepairs to 183diffcore mechanism as well as modified ones. This lets the copy 184detector consider unmodified files as copy source candidates at 185the expense of making it slower. Without `\--find-copies-harder`, 186'git-diff-{asterisk}' commands can detect copies only if the file that was 187copied happened to have been modified in the same changeset. 188 189 190diffcore-merge-broken: For Putting "Complete Rewrites" Back Together 191-------------------------------------------------------------------- 192 193This transformation is used to merge filepairs broken by 194diffcore-break, and not transformed into rename/copy by 195diffcore-rename, back into a single modification. This always 196runs when diffcore-break is used. 197 198For the purpose of merging broken filepairs back, it uses a 199different "extent of changes" computation from the ones used by 200diffcore-break and diffcore-rename. It counts only the deletion 201from the original, and does not count insertion. If you removed 202only 10 lines from a 100-line document, even if you added 910 203new lines to make a new 1000-line document, you did not do a 204complete rewrite. diffcore-break breaks such a case in order to 205help diffcore-rename to consider such filepairs as candidate of 206rename/copy detection, but if filepairs broken that way were not 207matched with other filepairs to create rename/copy, then this 208transformation merges them back into the original 209"modification". 210 211The "extent of changes" parameter can be tweaked from the 212default 80% (that is, unless more than 80% of the original 213material is deleted, the broken pairs are merged back into a 214single modification) by giving a second number to -B option, 215like these: 216 217* -B50/60 (give 50% "break score" to diffcore-break, use 60% 218 for diffcore-merge-broken). 219 220* -B/60 (the same as above, since diffcore-break defaults to 50%). 221 222Note that earlier implementation left a broken pair as a separate 223creation and deletion patches. This was an unnecessary hack and 224the latest implementation always merges all the broken pairs 225back into modifications, but the resulting patch output is 226formatted differently for easier review in case of such 227a complete rewrite by showing the entire contents of old version 228prefixed with '-', followed by the entire contents of new 229version prefixed with '+'. 230 231 232diffcore-pickaxe: For Detecting Addition/Deletion of Specified String 233--------------------------------------------------------------------- 234 235This transformation is used to find filepairs that represent 236changes that touch a specified string, and is controlled by the 237-S option and the `\--pickaxe-all` option to the 'git-diff-{asterisk}' 238commands. 239 240When diffcore-pickaxe is in use, it checks if there are 241filepairs whose "original" side has the specified string and 242whose "result" side does not. Such a filepair represents "the 243string appeared in this changeset". It also checks for the 244opposite case that loses the specified string. 245 246When `\--pickaxe-all` is not in effect, diffcore-pickaxe leaves 247only such filepairs that touch the specified string in its 248output. When `\--pickaxe-all` is used, diffcore-pickaxe leaves all 249filepairs intact if there is such a filepair, or makes the 250output empty otherwise. The latter behaviour is designed to 251make reviewing of the changes in the context of the whole 252changeset easier. 253 254 255diffcore-order: For Sorting the Output Based on Filenames 256--------------------------------------------------------- 257 258This is used to reorder the filepairs according to the user's 259(or project's) taste, and is controlled by the -O option to the 260'git-diff-{asterisk}' commands. 261 262This takes a text file each of whose lines is a shell glob 263pattern. Filepairs that match a glob pattern on an earlier line 264in the file are output before ones that match a later line, and 265filepairs that do not match any glob pattern are output last. 266 267As an example, a typical orderfile for the core git probably 268would look like this: 269 270------------------------------------------------ 271README 272Makefile 273Documentation 274*.h 275*.c 276t 277------------------------------------------------ 278 279SEE ALSO 280-------- 281linkgit:git-diff[1], 282linkgit:git-diff-files[1], 283linkgit:git-diff-index[1], 284linkgit:git-diff-tree[1], 285linkgit:git-format-patch[1], 286linkgit:git-log[1], 287linkgit:gitglossary[7], 288link:user-manual.html[The Git User's Manual] 289 290GIT 291--- 292Part of the linkgit:git[1] suite.