]>
Commit | Line | Data |
---|---|---|
8ac866a8 | 1 | //////////////////////////////////////////////////////////////// |
6ad6d3d3 | 2 | |
e83c5163 LT |
3 | GIT - the stupid content tracker |
4 | ||
8ac866a8 | 5 | //////////////////////////////////////////////////////////////// |
e83c5163 LT |
6 | "git" can mean anything, depending on your mood. |
7 | ||
8 | - random three-letter combination that is pronounceable, and not | |
9 | actually used by any common UNIX command. The fact that it is a | |
90c4851b | 10 | mispronunciation of "get" may or may not be relevant. |
e83c5163 LT |
11 | - stupid. contemptible and despicable. simple. Take your pick from the |
12 | dictionary of slang. | |
13 | - "global information tracker": you're in a good mood, and it actually | |
14 | works for you. Angels sing, and a light suddenly fills the room. | |
15 | - "goddamn idiotic truckload of sh*t": when it breaks | |
16 | ||
17 | This is a stupid (but extremely fast) directory content manager. It | |
8db9307c | 18 | doesn't do a whole lot, but what it 'does' do is track directory |
e83c5163 LT |
19 | contents efficiently. |
20 | ||
21 | There are two object abstractions: the "object database", and the | |
6ad6d3d3 LT |
22 | "current directory cache" aka "index". |
23 | ||
8ac866a8 DG |
24 | The Object Database |
25 | ~~~~~~~~~~~~~~~~~~~ | |
e83c5163 LT |
26 | The object database is literally just a content-addressable collection |
27 | of objects. All objects are named by their content, which is | |
28 | approximated by the SHA1 hash of the object itself. Objects may refer | |
8ac866a8 DG |
29 | to other objects (by referencing their SHA1 hash), and so you can |
30 | build up a hierarchy of objects. | |
e83c5163 | 31 | |
6ad6d3d3 LT |
32 | All objects have a statically determined "type" aka "tag", which is |
33 | determined at object creation time, and which identifies the format of | |
7096a645 | 34 | the object (i.e. how it is used, and how it can refer to other |
c4584ae3 JH |
35 | objects). There are currently four different object types: "blob", |
36 | "tree", "commit" and "tag". | |
6ad6d3d3 | 37 | |
2fa090b6 | 38 | A "blob" object cannot refer to any other object, and is, like the type |
6ad6d3d3 | 39 | implies, a pure storage object containing some user data. It is used to |
90c4851b | 40 | actually store the file data, i.e. a blob object is associated with some |
6ad6d3d3 LT |
41 | particular version of some file. |
42 | ||
43 | A "tree" object is an object that ties one or more "blob" objects into a | |
44 | directory structure. In addition, a tree object can refer to other tree | |
45 | objects, thus creating a directory hierarchy. | |
46 | ||
7096a645 | 47 | A "commit" object ties such directory hierarchies together into |
6ad6d3d3 LT |
48 | a DAG of revisions - each "commit" is associated with exactly one tree |
49 | (the directory hierarchy at the time of the commit). In addition, a | |
50 | "commit" refers to one or more "parent" commit objects that describe the | |
51 | history of how we arrived at that directory hierarchy. | |
52 | ||
53 | As a special case, a commit object with no parents is called the "root" | |
54 | object, and is the point of an initial project commit. Each project | |
55 | must have at least one root, and while you can tie several different | |
56 | root objects together into one project by creating a commit object which | |
57 | has two or more separate roots as its ultimate parents, that's probably | |
58 | just going to confuse people. So aim for the notion of "one root object | |
59 | per project", even if git itself does not enforce that. | |
60 | ||
8ac866a8 DG |
61 | A "tag" object symbolically identifies and can be used to sign other |
62 | objects. It contains the identifier and type of another object, a | |
63 | symbolic name (of course!) and, optionally, a signature. | |
64 | ||
2aef5bba DG |
65 | Regardless of object type, all objects share the following |
66 | characteristics: they are all deflated with zlib, and have a header | |
2fa090b6 | 67 | that not only specifies their type, but also provides size information |
2aef5bba | 68 | about the data in the object. It's worth noting that the SHA1 hash |
8db9307c JH |
69 | that is used to name the object is the hash of the original data |
70 | plus this header, so `sha1sum` 'file' does not match the object name | |
71 | for 'file'. | |
c4584ae3 | 72 | (Historical note: in the dawn of the age of git the hash |
8db9307c | 73 | was the sha1 of the 'compressed' object.) |
6ad6d3d3 LT |
74 | |
75 | As a result, the general consistency of an object can always be tested | |
e83c5163 LT |
76 | independently of the contents or the type of the object: all objects can |
77 | be validated by verifying that (a) their hashes match the content of the | |
78 | file and (b) the object successfully inflates to a stream of bytes that | |
2fa090b6 | 79 | forms a sequence of <ascii type without space> + <space> + <ascii decimal |
e83c5163 LT |
80 | size> + <byte\0> + <binary object data>. |
81 | ||
8ac866a8 DG |
82 | The structured objects can further have their structure and |
83 | connectivity to other objects verified. This is generally done with | |
215a7ad1 | 84 | the `git-fsck-objects` program, which generates a full dependency graph |
7096a645 DG |
85 | of all objects, and verifies their internal consistency (in addition |
86 | to just verifying their superficial consistency through the hash). | |
6ad6d3d3 LT |
87 | |
88 | The object types in some more detail: | |
89 | ||
8ac866a8 DG |
90 | Blob Object |
91 | ~~~~~~~~~~~ | |
92 | A "blob" object is nothing but a binary blob of data, and doesn't | |
93 | refer to anything else. There is no signature or any other | |
8db9307c | 94 | verification of the data, so while the object is consistent (it 'is' |
8ac866a8 DG |
95 | indexed by its sha1 hash, so the data itself is certainly correct), it |
96 | has absolutely no other attributes. No name associations, no | |
97 | permissions. It is purely a blob of data (i.e. normally "file | |
98 | contents"). | |
99 | ||
100 | In particular, since the blob is entirely defined by its data, if two | |
101 | files in a directory tree (or in multiple different versions of the | |
102 | repository) have the same contents, they will share the same blob | |
cdacb620 | 103 | object. The object is totally independent of its location in the |
8ac866a8 DG |
104 | directory tree, and renaming a file does not change the object that |
105 | file is associated with in any way. | |
106 | ||
a7154e91 SV |
107 | A blob is typically created when gitlink:git-update-index[1] |
108 | is run, and its data can be accessed by gitlink:git-cat-file[1]. | |
7096a645 | 109 | |
8ac866a8 DG |
110 | Tree Object |
111 | ~~~~~~~~~~~ | |
112 | The next hierarchical object type is the "tree" object. A tree object | |
113 | is a list of mode/name/blob data, sorted by name. Alternatively, the | |
114 | mode data may specify a directory mode, in which case instead of | |
115 | naming a blob, that name is associated with another TREE object. | |
116 | ||
117 | Like the "blob" object, a tree object is uniquely determined by the | |
118 | set contents, and so two separate but identical trees will always | |
119 | share the exact same object. This is true at all levels, i.e. it's | |
120 | true for a "leaf" tree (which does not refer to any other trees, only | |
121 | blobs) as well as for a whole subdirectory. | |
122 | ||
123 | For that reason a "tree" object is just a pure data abstraction: it | |
124 | has no history, no signatures, no verification of validity, except | |
125 | that since the contents are again protected by the hash itself, we can | |
126 | trust that the tree is immutable and its contents never change. | |
127 | ||
128 | So you can trust the contents of a tree to be valid, the same way you | |
129 | can trust the contents of a blob, but you don't know where those | |
8db9307c | 130 | contents 'came' from. |
8ac866a8 DG |
131 | |
132 | Side note on trees: since a "tree" object is a sorted list of | |
133 | "filename+content", you can create a diff between two trees without | |
134 | actually having to unpack two trees. Just ignore all common parts, | |
135 | and your diff will look right. In other words, you can effectively | |
136 | (and efficiently) tell the difference between any two random trees by | |
137 | O(n) where "n" is the size of the difference, rather than the size of | |
138 | the tree. | |
139 | ||
140 | Side note 2 on trees: since the name of a "blob" depends entirely and | |
141 | exclusively on its contents (i.e. there are no names or permissions | |
142 | involved), you can see trivial renames or permission changes by | |
143 | noticing that the blob stayed the same. However, renames with data | |
144 | changes need a smarter "diff" implementation. | |
145 | ||
a7154e91 SV |
146 | A tree is created with gitlink:git-write-tree[1] and |
147 | its data can be accessed by gitlink:git-ls-tree[1]. | |
148 | Two trees can be compared with gitlink:git-diff-tree[1]. | |
8ac866a8 | 149 | |
7096a645 DG |
150 | Commit Object |
151 | ~~~~~~~~~~~~~ | |
152 | The "commit" object is an object that introduces the notion of | |
8ac866a8 DG |
153 | history into the picture. In contrast to the other objects, it |
154 | doesn't just describe the physical state of a tree, it describes how | |
155 | we got there, and why. | |
156 | ||
7096a645 DG |
157 | A "commit" is defined by the tree-object that it results in, the |
158 | parent commits (zero, one or more) that led up to that point, and a | |
159 | comment on what happened. Again, a commit is not trusted per se: | |
8ac866a8 DG |
160 | the contents are well-defined and "safe" due to the cryptographically |
161 | strong signatures at all levels, but there is no reason to believe | |
162 | that the tree is "good" or that the merge information makes sense. | |
163 | The parents do not have to actually have any relationship with the | |
164 | result, for example. | |
165 | ||
7096a645 | 166 | Note on commits: unlike real SCM's, commits do not contain |
8db9307c | 167 | rename information or file mode change information. All of that is |
8ac866a8 DG |
168 | implicit in the trees involved (the result tree, and the result trees |
169 | of the parents), and describing that makes no sense in this idiotic | |
170 | file manager. | |
171 | ||
a7154e91 SV |
172 | A commit is created with gitlink:git-commit-tree[1] and |
173 | its data can be accessed by gitlink:git-cat-file[1]. | |
7096a645 DG |
174 | |
175 | Trust | |
176 | ~~~~~ | |
177 | An aside on the notion of "trust". Trust is really outside the scope | |
178 | of "git", but it's worth noting a few things. First off, since | |
8db9307c | 179 | everything is hashed with SHA1, you 'can' trust that an object is |
7096a645 DG |
180 | intact and has not been messed with by external sources. So the name |
181 | of an object uniquely identifies a known state - just not a state that | |
182 | you may want to trust. | |
8ac866a8 | 183 | |
7096a645 | 184 | Furthermore, since the SHA1 signature of a commit refers to the |
8ac866a8 | 185 | SHA1 signatures of the tree it is associated with and the signatures |
7096a645 | 186 | of the parent, a single named commit specifies uniquely a whole set |
8ac866a8 | 187 | of history, with full contents. You can't later fake any step of the |
7096a645 | 188 | way once you have the name of a commit. |
8ac866a8 DG |
189 | |
190 | So to introduce some real trust in the system, the only thing you need | |
8db9307c | 191 | to do is to digitally sign just 'one' special note, which includes the |
7096a645 DG |
192 | name of a top-level commit. Your digital signature shows others |
193 | that you trust that commit, and the immutability of the history of | |
194 | commits tells others that they can trust the whole history. | |
8ac866a8 DG |
195 | |
196 | In other words, you can easily validate a whole archive by just | |
197 | sending out a single email that tells the people the name (SHA1 hash) | |
7096a645 | 198 | of the top commit, and digitally sign that email using something |
8ac866a8 DG |
199 | like GPG/PGP. |
200 | ||
7096a645 | 201 | To assist in this, git also provides the tag object... |
8ac866a8 | 202 | |
7096a645 DG |
203 | Tag Object |
204 | ~~~~~~~~~~ | |
205 | Git provides the "tag" object to simplify creating, managing and | |
206 | exchanging symbolic and signed tokens. The "tag" object at its | |
207 | simplest simply symbolically identifies another object by containing | |
208 | the sha1, type and symbolic name. | |
8ac866a8 | 209 | |
7096a645 DG |
210 | However it can optionally contain additional signature information |
211 | (which git doesn't care about as long as there's less than 8k of | |
212 | it). This can then be verified externally to git. | |
8ac866a8 | 213 | |
7096a645 DG |
214 | Note that despite the tag features, "git" itself only handles content |
215 | integrity; the trust framework (and signature provision and | |
216 | verification) has to come from outside. | |
8ac866a8 | 217 | |
a7154e91 SV |
218 | A tag is created with gitlink:git-mktag[1], |
219 | its data can be accessed by gitlink:git-cat-file[1], | |
8db9307c | 220 | and the signature can be verified by |
a7154e91 | 221 | gitlink:git-verify-tag[1]. |
8ac866a8 | 222 | |
2aef5bba | 223 | |
8ac866a8 DG |
224 | The "index" aka "Current Directory Cache" |
225 | ----------------------------------------- | |
6ad6d3d3 LT |
226 | The index is a simple binary file, which contains an efficient |
227 | representation of a virtual directory content at some random time. It | |
228 | does so by a simple array that associates a set of names, dates, | |
229 | permissions and content (aka "blob") objects together. The cache is | |
230 | always kept ordered by name, and names are unique (with a few very | |
231 | specific rules) at any point in time, but the cache has no long-term | |
8ac866a8 | 232 | meaning, and can be partially updated at any time. |
6ad6d3d3 LT |
233 | |
234 | In particular, the index certainly does not need to be consistent with | |
235 | the current directory contents (in fact, most operations will depend on | |
8db9307c | 236 | different ways to make the index 'not' be consistent with the directory |
6ad6d3d3 | 237 | hierarchy), but it has three very important attributes: |
e83c5163 | 238 | |
8ac866a8 DG |
239 | '(a) it can re-generate the full state it caches (not just the |
240 | directory structure: it contains pointers to the "blob" objects so | |
241 | that it can regenerate the data too)' | |
e83c5163 | 242 | |
8ac866a8 DG |
243 | As a special case, there is a clear and unambiguous one-way mapping |
244 | from a current directory cache to a "tree object", which can be | |
245 | efficiently created from just the current directory cache without | |
246 | actually looking at any other data. So a directory cache at any one | |
247 | time uniquely specifies one and only one "tree" object (but has | |
248 | additional data to make it easy to match up that tree object with what | |
249 | has happened in the directory) | |
e83c5163 | 250 | |
8ac866a8 DG |
251 | '(b) it has efficient methods for finding inconsistencies between that |
252 | cached state ("tree object waiting to be instantiated") and the | |
253 | current state.' | |
e83c5163 | 254 | |
8ac866a8 DG |
255 | '(c) it can additionally efficiently represent information about merge |
256 | conflicts between different tree objects, allowing each pathname to be | |
257 | associated with sufficient information about the trees involved that | |
258 | you can create a three-way merge between them.' | |
6ad6d3d3 LT |
259 | |
260 | Those are the three ONLY things that the directory cache does. It's a | |
e83c5163 LT |
261 | cache, and the normal operation is to re-generate it completely from a |
262 | known tree object, or update/compare it with a live tree that is being | |
6ad6d3d3 LT |
263 | developed. If you blow the directory cache away entirely, you generally |
264 | haven't lost any information as long as you have the name of the tree | |
265 | that it described. | |
266 | ||
8db9307c | 267 | At the same time, the index is at the same time also the |
6ad6d3d3 LT |
268 | staging area for creating new trees, and creating a new tree always |
269 | involves a controlled modification of the index file. In particular, | |
270 | the index file can have the representation of an intermediate tree that | |
271 | has not yet been instantiated. So the index can be thought of as a | |
272 | write-back cache, which can contain dirty information that has not yet | |
8ac866a8 | 273 | been written back to the backing store. |
6ad6d3d3 LT |
274 | |
275 | ||
276 | ||
8ac866a8 DG |
277 | The Workflow |
278 | ------------ | |
6ad6d3d3 | 279 | Generally, all "git" operations work on the index file. Some operations |
8ac866a8 | 280 | work *purely* on the index file (showing the current state of the |
6ad6d3d3 LT |
281 | index), but most operations move data to and from the index file. Either |
282 | from the database or from the working directory. Thus there are four | |
283 | main combinations: | |
284 | ||
8ac866a8 DG |
285 | 1) working directory -> index |
286 | ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ | |
6ad6d3d3 | 287 | |
8ac866a8 | 288 | You update the index with information from the working directory with |
a7154e91 | 289 | the gitlink:git-update-index[1] command. You |
7096a645 DG |
290 | generally update the index information by just specifying the filename |
291 | you want to update, like so: | |
6ad6d3d3 | 292 | |
215a7ad1 | 293 | git-update-index filename |
6ad6d3d3 | 294 | |
8ac866a8 DG |
295 | but to avoid common mistakes with filename globbing etc, the command |
296 | will not normally add totally new entries or remove old entries, | |
297 | i.e. it will normally just update existing cache entries. | |
6ad6d3d3 | 298 | |
8ac866a8 | 299 | To tell git that yes, you really do realize that certain files no |
2fa090b6 | 300 | longer exist, or that new files should be added, you |
8db9307c | 301 | should use the `--remove` and `--add` flags respectively. |
6ad6d3d3 | 302 | |
8db9307c | 303 | NOTE! A `--remove` flag does 'not' mean that subsequent filenames will |
8ac866a8 DG |
304 | necessarily be removed: if the files still exist in your directory |
305 | structure, the index will be updated with their new status, not | |
8db9307c | 306 | removed. The only thing `--remove` means is that update-cache will be |
8ac866a8 DG |
307 | considering a removed file to be a valid thing, and if the file really |
308 | does not exist any more, it will update the index accordingly. | |
6ad6d3d3 | 309 | |
215a7ad1 | 310 | As a special case, you can also do `git-update-index --refresh`, which |
8ac866a8 | 311 | will refresh the "stat" information of each index to match the current |
8db9307c | 312 | stat information. It will 'not' update the object status itself, and |
8ac866a8 DG |
313 | it will only update the fields that are used to quickly test whether |
314 | an object still matches its old backing store object. | |
6ad6d3d3 | 315 | |
8ac866a8 DG |
316 | 2) index -> object database |
317 | ~~~~~~~~~~~~~~~~~~~~~~~~~~~ | |
6ad6d3d3 | 318 | |
8ac866a8 | 319 | You write your current index file to a "tree" object with the program |
6ad6d3d3 | 320 | |
7096a645 | 321 | git-write-tree |
6ad6d3d3 | 322 | |
8ac866a8 DG |
323 | that doesn't come with any options - it will just write out the |
324 | current index into the set of tree objects that describe that state, | |
325 | and it will return the name of the resulting top-level tree. You can | |
326 | use that tree to re-generate the index at any time by going in the | |
327 | other direction: | |
6ad6d3d3 | 328 | |
8ac866a8 DG |
329 | 3) object database -> index |
330 | ~~~~~~~~~~~~~~~~~~~~~~~~~~~ | |
6ad6d3d3 | 331 | |
8ac866a8 DG |
332 | You read a "tree" file from the object database, and use that to |
333 | populate (and overwrite - don't do this if your index contains any | |
334 | unsaved state that you might want to restore later!) your current | |
335 | index. Normal operation is just | |
6ad6d3d3 | 336 | |
7096a645 | 337 | git-read-tree <sha1 of tree> |
6ad6d3d3 | 338 | |
8ac866a8 | 339 | and your index file will now be equivalent to the tree that you saved |
8db9307c | 340 | earlier. However, that is only your 'index' file: your working |
8ac866a8 | 341 | directory contents have not been modified. |
6ad6d3d3 | 342 | |
8ac866a8 DG |
343 | 4) index -> working directory |
344 | ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ | |
6ad6d3d3 | 345 | |
8ac866a8 DG |
346 | You update your working directory from the index by "checking out" |
347 | files. This is not a very common operation, since normally you'd just | |
348 | keep your files updated, and rather than write to your working | |
349 | directory, you'd tell the index files about the changes in your | |
215a7ad1 | 350 | working directory (i.e. `git-update-index`). |
6ad6d3d3 | 351 | |
8ac866a8 DG |
352 | However, if you decide to jump to a new version, or check out somebody |
353 | else's version, or just restore a previous tree, you'd populate your | |
354 | index file with read-tree, and then you need to check out the result | |
355 | with | |
8db9307c | 356 | |
215a7ad1 | 357 | git-checkout-index filename |
6ad6d3d3 | 358 | |
8db9307c | 359 | or, if you want to check out all of the index, use `-a`. |
6ad6d3d3 | 360 | |
215a7ad1 | 361 | NOTE! git-checkout-index normally refuses to overwrite old files, so |
7096a645 | 362 | if you have an old version of the tree already checked out, you will |
8db9307c JH |
363 | need to use the "-f" flag ('before' the "-a" flag or the filename) to |
364 | 'force' the checkout. | |
6ad6d3d3 LT |
365 | |
366 | ||
8ac866a8 DG |
367 | Finally, there are a few odds and ends which are not purely moving |
368 | from one representation to the other: | |
6ad6d3d3 | 369 | |
8ac866a8 DG |
370 | 5) Tying it all together |
371 | ~~~~~~~~~~~~~~~~~~~~~~~~ | |
7096a645 DG |
372 | To commit a tree you have instantiated with "git-write-tree", you'd |
373 | create a "commit" object that refers to that tree and the history | |
374 | behind it - most notably the "parent" commits that preceded it in | |
375 | history. | |
6ad6d3d3 | 376 | |
8ac866a8 DG |
377 | Normally a "commit" has one parent: the previous state of the tree |
378 | before a certain change was made. However, sometimes it can have two | |
379 | or more parent commits, in which case we call it a "merge", due to the | |
380 | fact that such a commit brings together ("merges") two or more | |
381 | previous states represented by other commits. | |
6ad6d3d3 | 382 | |
8ac866a8 DG |
383 | In other words, while a "tree" represents a particular directory state |
384 | of a working directory, a "commit" represents that state in "time", | |
385 | and explains how we got there. | |
6ad6d3d3 | 386 | |
8ac866a8 DG |
387 | You create a commit object by giving it the tree that describes the |
388 | state at the time of the commit, and a list of parents: | |
6ad6d3d3 | 389 | |
7096a645 | 390 | git-commit-tree <tree> -p <parent> [-p <parent2> ..] |
6ad6d3d3 | 391 | |
8ac866a8 DG |
392 | and then giving the reason for the commit on stdin (either through |
393 | redirection from a pipe or file, or by just typing it at the tty). | |
6ad6d3d3 | 394 | |
7096a645 DG |
395 | git-commit-tree will return the name of the object that represents |
396 | that commit, and you should save it away for later use. Normally, | |
8db9307c | 397 | you'd commit a new `HEAD` state, and while git doesn't care where you |
7096a645 | 398 | save the note about that state, in practice we tend to just write the |
cd0a781c JH |
399 | result to the file pointed at by `.git/HEAD`, so that we can always see |
400 | what the last committed state was. | |
6ad6d3d3 | 401 | |
66158e33 JH |
402 | Here is an ASCII art by Jon Loeliger that illustrates how |
403 | various pieces fit together. | |
404 | ||
405 | ------------ | |
406 | ||
407 | commit-tree | |
408 | commit obj | |
409 | +----+ | |
410 | | | | |
411 | | | | |
412 | V V | |
413 | +-----------+ | |
414 | | Object DB | | |
415 | | Backing | | |
416 | | Store | | |
417 | +-----------+ | |
418 | ^ | |
419 | write-tree | | | |
420 | tree obj | | | |
421 | | | read-tree | |
422 | | | tree obj | |
423 | V | |
424 | +-----------+ | |
425 | | Index | | |
426 | | "cache" | | |
427 | +-----------+ | |
428 | update-index ^ | |
429 | blob obj | | | |
430 | | | | |
431 | checkout-index -u | | checkout-index | |
432 | stat | | blob obj | |
433 | V | |
434 | +-----------+ | |
435 | | Working | | |
436 | | Directory | | |
437 | +-----------+ | |
438 | ||
439 | ------------ | |
440 | ||
441 | ||
8ac866a8 DG |
442 | 6) Examining the data |
443 | ~~~~~~~~~~~~~~~~~~~~~ | |
6ad6d3d3 | 444 | |
8ac866a8 DG |
445 | You can examine the data represented in the object database and the |
446 | index with various helper tools. For every object, you can use | |
a7154e91 | 447 | gitlink:git-cat-file[1] to examine details about the |
7096a645 | 448 | object: |
6ad6d3d3 | 449 | |
7096a645 | 450 | git-cat-file -t <objectname> |
6ad6d3d3 | 451 | |
8ac866a8 DG |
452 | shows the type of the object, and once you have the type (which is |
453 | usually implicit in where you find the object), you can use | |
6ad6d3d3 | 454 | |
8db9307c | 455 | git-cat-file blob|tree|commit|tag <objectname> |
6ad6d3d3 | 456 | |
8ac866a8 | 457 | to show its contents. NOTE! Trees have binary content, and as a result |
7096a645 | 458 | there is a special helper for showing that content, called |
8db9307c | 459 | `git-ls-tree`, which turns the binary content into a more easily |
7096a645 | 460 | readable form. |
6ad6d3d3 | 461 | |
8ac866a8 DG |
462 | It's especially instructive to look at "commit" objects, since those |
463 | tend to be small and fairly self-explanatory. In particular, if you | |
8db9307c | 464 | follow the convention of having the top commit name in `.git/HEAD`, |
8ac866a8 | 465 | you can do |
6ad6d3d3 | 466 | |
cd0a781c | 467 | git-cat-file commit HEAD |
6ad6d3d3 | 468 | |
8ac866a8 | 469 | to see what the top commit was. |
6ad6d3d3 | 470 | |
8ac866a8 DG |
471 | 7) Merging multiple trees |
472 | ~~~~~~~~~~~~~~~~~~~~~~~~~ | |
6ad6d3d3 | 473 | |
8ac866a8 DG |
474 | Git helps you do a three-way merge, which you can expand to n-way by |
475 | repeating the merge procedure arbitrary times until you finally | |
476 | "commit" the state. The normal situation is that you'd only do one | |
477 | three-way merge (two parents), and commit it, but if you like to, you | |
478 | can do multiple parents in one go. | |
6ad6d3d3 | 479 | |
8ac866a8 DG |
480 | To do a three-way merge, you need the two sets of "commit" objects |
481 | that you want to merge, use those to find the closest common parent (a | |
482 | third "commit" object), and then use those commit objects to find the | |
483 | state of the directory ("tree" object) at these points. | |
6ad6d3d3 | 484 | |
8ac866a8 DG |
485 | To get the "base" for the merge, you first look up the common parent |
486 | of two commits with | |
6ad6d3d3 | 487 | |
7096a645 | 488 | git-merge-base <commit1> <commit2> |
6ad6d3d3 | 489 | |
8ac866a8 DG |
490 | which will return you the commit they are both based on. You should |
491 | now look up the "tree" objects of those commits, which you can easily | |
492 | do with (for example) | |
6ad6d3d3 | 493 | |
7096a645 | 494 | git-cat-file commit <commitname> | head -1 |
6ad6d3d3 | 495 | |
8ac866a8 DG |
496 | since the tree object information is always the first line in a commit |
497 | object. | |
498 | ||
499 | Once you know the three trees you are going to merge (the one | |
500 | "original" tree, aka the common case, and the two "result" trees, aka | |
501 | the branches you want to merge), you do a "merge" read into the | |
8db9307c | 502 | index. This will complain if it has to throw away your old index contents, so you should |
8ac866a8 DG |
503 | make sure that you've committed those - in fact you would normally |
504 | always do a merge against your last commit (which should thus match | |
505 | what you have in your current index anyway). | |
6ad6d3d3 | 506 | |
8ac866a8 | 507 | To do the merge, do |
6ad6d3d3 | 508 | |
8db9307c | 509 | git-read-tree -m -u <origtree> <yourtree> <targettree> |
6ad6d3d3 | 510 | |
8ac866a8 | 511 | which will do all trivial merge operations for you directly in the |
7096a645 | 512 | index file, and you can just write the result out with |
8db9307c JH |
513 | `git-write-tree`. |
514 | ||
515 | Historical note. We did not have `-u` facility when this | |
516 | section was first written, so we used to warn that | |
517 | the merge is done in the index file, not in your | |
2fa090b6 JH |
518 | working tree, and your working tree will not match your |
519 | index after this step. | |
520 | This is no longer true. The above command, thanks to `-u` | |
521 | option, updates your working tree with the merge results for | |
522 | paths that have been trivially merged. | |
6ad6d3d3 | 523 | |
6ad6d3d3 | 524 | |
8db9307c JH |
525 | 8) Merging multiple trees, continued |
526 | ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ | |
527 | ||
528 | Sadly, many merges aren't trivial. If there are files that have | |
8ac866a8 DG |
529 | been added.moved or removed, or if both branches have modified the |
530 | same file, you will be left with an index tree that contains "merge | |
8db9307c | 531 | entries" in it. Such an index tree can 'NOT' be written out to a tree |
8ac866a8 DG |
532 | object, and you will have to resolve any such merge clashes using |
533 | other tools before you can write out the result. | |
6ad6d3d3 | 534 | |
8db9307c JH |
535 | You can examine such index state with `git-ls-files --unmerged` |
536 | command. An example: | |
537 | ||
538 | ------------------------------------------------ | |
539 | $ git-read-tree -m $orig HEAD $target | |
540 | $ git-ls-files --unmerged | |
541 | 100644 263414f423d0e4d70dae8fe53fa34614ff3e2860 1 hello.c | |
542 | 100644 06fa6a24256dc7e560efa5687fa84b51f0263c3a 2 hello.c | |
543 | 100644 cc44c73eb783565da5831b4d820c962954019b69 3 hello.c | |
544 | ------------------------------------------------ | |
545 | ||
546 | Each line of the `git-ls-files --unmerged` output begins with | |
547 | the blob mode bits, blob SHA1, 'stage number', and the | |
548 | filename. The 'stage number' is git's way to say which tree it | |
549 | came from: stage 1 corresponds to `$orig` tree, stage 2 `HEAD` | |
550 | tree, and stage3 `$target` tree. | |
551 | ||
552 | Earlier we said that trivial merges are done inside | |
553 | `git-read-tree -m`. For example, if the file did not change | |
554 | from `$orig` to `HEAD` nor `$target`, or if the file changed | |
555 | from `$orig` to `HEAD` and `$orig` to `$target` the same way, | |
556 | obviously the final outcome is what is in `HEAD`. What the | |
557 | above example shows is that file `hello.c` was changed from | |
558 | `$orig` to `HEAD` and `$orig` to `$target` in a different way. | |
559 | You could resolve this by running your favorite 3-way merge | |
560 | program, e.g. `diff3` or `merge`, on the blob objects from | |
561 | these three stages yourself, like this: | |
562 | ||
563 | ------------------------------------------------ | |
564 | $ git-cat-file blob 263414f... >hello.c~1 | |
565 | $ git-cat-file blob 06fa6a2... >hello.c~2 | |
566 | $ git-cat-file blob cc44c73... >hello.c~3 | |
567 | $ merge hello.c~2 hello.c~1 hello.c~3 | |
568 | ------------------------------------------------ | |
569 | ||
570 | This would leave the merge result in `hello.c~2` file, along | |
571 | with conflict markers if there are conflicts. After verifying | |
572 | the merge result makes sense, you can tell git what the final | |
573 | merge result for this file is by: | |
574 | ||
575 | mv -f hello.c~2 hello.c | |
215a7ad1 | 576 | git-update-index hello.c |
8db9307c | 577 | |
215a7ad1 | 578 | When a path is in unmerged state, running `git-update-index` for |
8db9307c JH |
579 | that path tells git to mark the path resolved. |
580 | ||
581 | The above is the description of a git merge at the lowest level, | |
582 | to help you understand what conceptually happens under the hood. | |
583 | In practice, nobody, not even git itself, uses three `git-cat-file` | |
215a7ad1 | 584 | for this. There is `git-merge-index` program that extracts the |
2fa090b6 | 585 | stages to temporary files and calls a "merge" script on it: |
8db9307c | 586 | |
215a7ad1 | 587 | git-merge-index git-merge-one-file hello.c |
8db9307c JH |
588 | |
589 | and that is what higher level `git resolve` is implemented with. |