Documentation/revisions.txt

   1 SPECIFYING REVISIONS
   2 --------------------
   3
   4 A revision parameter typically, but not necessarily, names a
   5 commit object.  They use what is called an 'extended SHA1'
   6 syntax.  Here are various ways to spell object names.  The
   7 ones listed near the end of this list are to name trees and
   8 blobs contained in a commit.
   9
  10 * The full SHA1 object name (40-byte hexadecimal string), or
  11   a substring of such that is unique within the repository.
  12   E.g. dae86e1950b1277e545cee180551750029cfe735 and dae86e both
  13   name the same commit object if there are no other object in
  14   your repository whose object name starts with dae86e.
  15
  16 * An output from 'git describe'; i.e. a closest tag, optionally
  17   followed by a dash and a number of commits, followed by a dash, a
  18   `g`, and an abbreviated object name.
  19
  20 * A symbolic ref name.  E.g. 'master' typically means the commit
  21   object referenced by refs/heads/master.  If you
  22   happen to have both heads/master and tags/master, you can
  23   explicitly say 'heads/master' to tell git which one you mean.
  24   When ambiguous, a `<name>` is disambiguated by taking the
  25   first match in the following rules:
  26
  27   . if `$GIT_DIR/<name>` exists, that is what you mean (this is usually
  28     useful only for `HEAD`, `FETCH_HEAD`, `ORIG_HEAD` and `MERGE_HEAD`);
  29
  30   . otherwise, `refs/<name>` if exists;
  31
  32   . otherwise, `refs/tags/<name>` if exists;
  33
  34   . otherwise, `refs/heads/<name>` if exists;
  35
  36   . otherwise, `refs/remotes/<name>` if exists;
  37
  38   . otherwise, `refs/remotes/<name>/HEAD` if exists.
  39 +
  40 HEAD names the commit your changes in the working tree is based on.
  41 FETCH_HEAD records the branch you fetched from a remote repository
  42 with your last 'git fetch' invocation.
  43 ORIG_HEAD is created by commands that moves your HEAD in a drastic
  44 way, to record the position of the HEAD before their operation, so that
  45 you can change the tip of the branch back to the state before you ran
  46 them easily.
  47 MERGE_HEAD records the commit(s) you are merging into your branch
  48 when you run 'git merge'.
  49 +
  50 Note that any of the `refs/*` cases above may come either from
  51 the `$GIT_DIR/refs` directory or from the `$GIT_DIR/packed-refs` file.
  52
  53 * A ref followed by the suffix '@' with a date specification
  54   enclosed in a brace
  55   pair (e.g. '\{yesterday\}', '\{1 month 2 weeks 3 days 1 hour 1
  56   second ago\}' or '\{1979-02-26 18:30:00\}') to specify the value
  57   of the ref at a prior point in time.  This suffix may only be
  58   used immediately following a ref name and the ref must have an
  59   existing log ($GIT_DIR/logs/<ref>). Note that this looks up the state
  60   of your *local* ref at a given time; e.g., what was in your local
  61   `master` branch last week. If you want to look at commits made during
  62   certain times, see `--since` and `--until`.
  63
  64 * A ref followed by the suffix '@' with an ordinal specification
  65   enclosed in a brace pair (e.g. '\{1\}', '\{15\}') to specify
  66   the n-th prior value of that ref.  For example 'master@\{1\}'
  67   is the immediate prior value of 'master' while 'master@\{5\}'
  68   is the 5th prior value of 'master'. This suffix may only be used
  69   immediately following a ref name and the ref must have an existing
  70   log ($GIT_DIR/logs/<ref>).
  71
  72 * You can use the '@' construct with an empty ref part to get at a
  73   reflog of the current branch. For example, if you are on the
  74   branch 'blabla', then '@\{1\}' means the same as 'blabla@\{1\}'.
  75
  76 * The special construct '@\{-<n>\}' means the <n>th branch checked out
  77   before the current one.
  78
  79 * The suffix '@\{upstream\}' to a ref (short form 'ref@\{u\}') refers to
  80   the branch the ref is set to build on top of.  Missing ref defaults
  81   to the current branch.
  82
  83 * A suffix '{caret}' to a revision parameter (e.g. 'HEAD{caret}') means the first parent of
  84   that commit object.  '{caret}<n>' means the <n>th parent (i.e.
  85   'rev{caret}'
  86   is equivalent to 'rev{caret}1').  As a special rule,
  87   'rev{caret}0' means the commit itself and is used when 'rev' is the
  88   object name of a tag object that refers to a commit object.
  89
  90 * A suffix '{tilde}<n>' to a revision parameter means the commit
  91   object that is the <n>th generation grand-parent of the named
  92   commit object, following only the first parent.  I.e. rev~3 is
  93   equivalent to rev{caret}{caret}{caret} which is equivalent to
  94   rev{caret}1{caret}1{caret}1.  See below for a illustration of
  95   the usage of this form.
  96
  97 * A suffix '{caret}' followed by an object type name enclosed in
  98   brace pair (e.g. `v0.99.8{caret}\{commit\}`) means the object
  99   could be a tag, and dereference the tag recursively until an
 100   object of that type is found or the object cannot be
 101   dereferenced anymore (in which case, barf).  `rev{caret}0`
 102   introduced earlier is a short-hand for `rev{caret}\{commit\}`.
 103
 104 * A suffix '{caret}' followed by an empty brace pair
 105   (e.g. `v0.99.8{caret}\{\}`) means the object could be a tag,
 106   and dereference the tag recursively until a non-tag object is
 107   found.
 108
 109 * A colon, followed by a slash, followed by a text (e.g. `:/fix nasty bug`): this names
 110   a commit whose commit message matches the specified regular expression.
 111   This name returns the youngest matching commit which is
 112   reachable from any ref.  If the commit message starts with a
 113   '!', you have to repeat that;  the special sequence ':/!',
 114   followed by something else than '!' is reserved for now.
 115   The regular expression can match any part of the commit message. To
 116   match messages starting with a string, one can use e.g. `:/^foo`.
 117
 118 * A suffix ':' followed by a path (e.g. `HEAD:README`); this names the blob or tree
 119   at the given path in the tree-ish object named by the part
 120   before the colon.
 121   ':path' (with an empty part before the colon, e.g. `:README`)
 122   is a special case of the syntax described next: content
 123   recorded in the index at the given path.
 124
 125 * A colon, optionally followed by a stage number (0 to 3) and a
 126   colon, followed by a path (e.g. `:0:README`); this names a blob object in the
 127   index at the given path. Missing stage number (and the colon
 128   that follows it, e.g. `:README`) names a stage 0 entry. During a merge, stage
 129   1 is the common ancestor, stage 2 is the target branch's version
 130   (typically the current branch), and stage 3 is the version from
 131   the branch being merged.
 132
 133 Here is an illustration, by Jon Loeliger.  Both commit nodes B
 134 and C are parents of commit node A.  Parent commits are ordered
 135 left-to-right.
 136
 137 ........................................
 138 G   H   I   J
 139  \ /     \ /
 140   D   E   F
 141    \  |  / \
 142     \ | /   |
 143      \|/    |
 144       B     C
 145        \   /
 146         \ /
 147          A
 148 ........................................
 149
 150     A =      = A^0
 151     B = A^   = A^1     = A~1
 152     C = A^2  = A^2
 153     D = A^^  = A^1^1   = A~2
 154     E = B^2  = A^^2
 155     F = B^3  = A^^3
 156     G = A^^^ = A^1^1^1 = A~3
 157     H = D^2  = B^^2    = A^^^2  = A~2^2
 158     I = F^   = B^3^    = A^^3^
 159     J = F^2  = B^3^2   = A^^3^2
 160
 161
 162 SPECIFYING RANGES
 163 -----------------
 164
 165 History traversing commands such as 'git log' operate on a set
 166 of commits, not just a single commit.  To these commands,
 167 specifying a single revision with the notation described in the
 168 previous section means the set of commits reachable from that
 169 commit, following the commit ancestry chain.
 170
 171 To exclude commits reachable from a commit, a prefix `{caret}`
 172 notation is used.  E.g. `{caret}r1 r2` means commits reachable
 173 from `r2` but exclude the ones reachable from `r1`.
 174
 175 This set operation appears so often that there is a shorthand
 176 for it.  When you have two commits `r1` and `r2` (named according
 177 to the syntax explained in SPECIFYING REVISIONS above), you can ask
 178 for commits that are reachable from r2 excluding those that are reachable
 179 from r1 by `{caret}r1 r2` and it can be written as `r1..r2`.
 180
 181 A similar notation `r1\...r2` is called symmetric difference
 182 of `r1` and `r2` and is defined as
 183 `r1 r2 --not $(git merge-base --all r1 r2)`.
 184 It is the set of commits that are reachable from either one of
 185 `r1` or `r2` but not from both.
 186
 187 Two other shorthands for naming a set that is formed by a commit
 188 and its parent commits exist.  The `r1{caret}@` notation means all
 189 parents of `r1`.  `r1{caret}!` includes commit `r1` but excludes
 190 all of its parents.
 191
 192 Here are a handful of examples:
 193
 194    D                G H D
 195    D F              G H I J D F
 196    ^G D             H D
 197    ^D B             E I J F B
 198    B...C            G H D E B C
 199    ^D B C           E I J F B C
 200    C^@              I J F
 201    F^! D            G H D F