1A tutorial introduction to git: part two 2======================================== 3 4You should work through link:tutorial.html[A tutorial introduction to 5git] before reading this tutorial. 6 7The goal of this tutorial is to introduce two fundamental pieces of 8git's architecture--the object database and the index file--and to 9provide the reader with everything necessary to understand the rest 10of the git documentation. 11 12The git object database 13----------------------- 14 15Let's start a new project and create a small amount of history: 16 17------------------------------------------------ 18$ mkdir test-project 19$ cd test-project 20$ git init-db 21defaulting to local storage area 22$ echo 'hello world' > file.txt 23$ git add . 24$ git commit -a -m "initial commit" 25Committing initial tree 92b8b694ffb1675e5975148e1121810081dbdffe 26$ echo 'hello world!' >file.txt 27$ git commit -a -m "add emphasis" 28------------------------------------------------ 29 30What are the 40 digits of hex that git responded to the first commit 31with? 32 33We saw in part one of the tutorial that commits have names like this. 34It turns out that every object in the git history is stored under 35such a 40-digit hex name. That name is the SHA1 hash of the object's 36contents; among other things, this ensures that git will never store 37the same data twice (since identical data is given an identical SHA1 38name), and that the contents of a git object will never change (since 39that would change the object's name as well). 40 41We can ask git about this particular object with the cat-file 42command--just cut-and-paste from the reply to the initial commit, to 43save yourself typing all 40 hex digits: 44 45------------------------------------------------ 46$ git cat-file -t 92b8b694ffb1675e5975148e1121810081dbdffe 47tree 48------------------------------------------------ 49 50A tree can refer to one or more "blob" objects, each corresponding to 51a file. In addition, a tree can also refer to other tree objects, 52thus creating a directory heirarchy. You can examine the contents of 53any tree using ls-tree (remember that a long enough initial portion 54of the SHA1 will also work): 55 56------------------------------------------------ 57$ git ls-tree 92b8b694 58100644 blob 3b18e512dba79e4c8300dd08aeb37f8e728b8dad file.txt 59------------------------------------------------ 60 61Thus we see that this tree has one file in it. The SHA1 hash is a 62reference to that file's data: 63 64------------------------------------------------ 65$ git cat-file -t 3b18e512 66blob 67------------------------------------------------ 68 69A "blob" is just file data, which we can also examine with cat-file: 70 71------------------------------------------------ 72$ git cat-file blob 3b18e512 73hello world 74------------------------------------------------ 75 76Note that this is the old file data; so the object that git named in 77its response to the initial tree was a tree with a snapshot of the 78directory state that was recorded by the first commit. 79 80All of these objects are stored under their SHA1 names inside the git 81directory: 82 83------------------------------------------------ 84$ find .git/objects/ 85.git/objects/ 86.git/objects/pack 87.git/objects/info 88.git/objects/3b 89.git/objects/3b/18e512dba79e4c8300dd08aeb37f8e728b8dad 90.git/objects/92 91.git/objects/92/b8b694ffb1675e5975148e1121810081dbdffe 92.git/objects/54 93.git/objects/54/196cc2703dc165cbd373a65a4dcf22d50ae7f7 94.git/objects/a0 95.git/objects/a0/423896973644771497bdc03eb99d5281615b51 96.git/objects/d0 97.git/objects/d0/492b368b66bdabf2ac1fd8c92b39d3db916e59 98.git/objects/c4 99.git/objects/c4/d59f390b9cfd4318117afde11d601c1085f241 100------------------------------------------------ 101 102and the contents of these files is just the compressed data plus a 103header identifying their length and their type. The type is either a 104blob, a tree, a commit, or a tag. We've seen a blob and a tree now, 105so next we should look at a commit. 106 107The simplest commit to find is the HEAD commit, which we can find 108from .git/HEAD: 109 110------------------------------------------------ 111$ cat .git/HEAD 112ref: refs/heads/master 113------------------------------------------------ 114 115As you can see, this tells us which branch we're currently on, and it 116tells us this by naming a file under the .git directory, which itself 117contains a SHA1 name referring to a commit object, which we can 118examine with cat-file: 119 120------------------------------------------------ 121$ cat .git/refs/heads/master 122c4d59f390b9cfd4318117afde11d601c1085f241 123$ git cat-file -t c4d59f39 124commit 125$ git cat-file commit c4d59f39 126tree d0492b368b66bdabf2ac1fd8c92b39d3db916e59 127parent 54196cc2703dc165cbd373a65a4dcf22d50ae7f7 128author J. Bruce Fields <bfields@puzzle.fieldses.org> 1143418702 -0500 129committer J. Bruce Fields <bfields@puzzle.fieldses.org> 1143418702 -0500 130 131add emphasis 132------------------------------------------------ 133 134The "tree" object here refers to the new state of the tree: 135 136------------------------------------------------ 137$ git ls-tree d0492b36 138100644 blob a0423896973644771497bdc03eb99d5281615b51 file.txt 139$ git cat-file commit a0423896 140hello world! 141------------------------------------------------ 142 143and the "parent" object refers to the previous commit: 144 145------------------------------------------------ 146$ git-cat-file commit 54196cc2 147tree 92b8b694ffb1675e5975148e1121810081dbdffe 148author J. Bruce Fields <bfields@puzzle.fieldses.org> 1143414668 -0500 149committer J. Bruce Fields <bfields@puzzle.fieldses.org> 1143414668 -0500 150 151initial commit 152------------------------------------------------ 153 154The tree object is the tree we examined first, and this commit is 155unusual in that it lacks any parent. 156 157Most commits have only one parent, but it is also common for a commit 158to have multiple parents. In that case the commit represents a 159merge, with the parent references pointing to the heads of the merged 160branches. 161 162Besides blobs, trees, and commits, the only remaining type of object 163is a "tag", which we won't discuss here; refer to gitlink:git-tag[1] 164for details. 165 166So now we know how git uses the object database to represent a 167project's history: 168 169 * "commit" objects refer to "tree" objects representing the 170 snapshot of a directory tree at a particular point in the 171 history, and refer to "parent" commits to show how they're 172 connected into the project history. 173 * "tree" objects represent the state of a single directory, 174 associating directory names to "blob" objects containing file 175 data and "tree" objects containing subdirectory information. 176 * "blob" objects contain file data without any other structure. 177 * References to commit objects at the head of each branch are 178 stored in files under .git/refs/heads/. 179 * The name of the current branch is stored in .git/HEAD. 180 181Note, by the way, that lots of commands take a tree as an argument. 182But as we can see above, a tree can be referred to in many different 183ways--by the SHA1 name for that tree, by the name of a commit that 184refers to the tree, by the name of a branch whose head refers to that 185tree, etc.--and most such commands can accept any of these names. 186 187In command synopses, the word "tree-ish" is sometimes used to 188designate such an argument. 189 190The index file 191-------------- 192 193The primary tool we've been using to create commits is "git commit 194-a", which creates a commit including every change you've made to 195your working tree. But what if you want to commit changes only to 196certain files? Or only certain changes to certain files? 197 198If we look at the way commits are created under the cover, we'll see 199that there are more flexible ways creating commits. 200 201Continuing with our test-project, let's modify file.txt again: 202 203------------------------------------------------ 204$ echo "hello world, again" >>file.txt 205------------------------------------------------ 206 207but this time instead of immediately making the commit, let's take an 208intermediate step, and ask for diffs along the way to keep track of 209what's happening: 210 211------------------------------------------------ 212$ git diff 213--- a/file.txt 214+++ b/file.txt 215@@ -1 +1,2 @@ 216 hello world! 217+hello world, again 218$ git update-index file.txt 219$ git diff 220------------------------------------------------ 221 222The last diff is empty, but no new commits have been made, and the 223head still doesn't contain the new line: 224 225------------------------------------------------ 226$ git-diff HEAD 227diff --git a/file.txt b/file.txt 228index a042389..513feba 100644 229--- a/file.txt 230+++ b/file.txt 231@@ -1 +1,2 @@ 232 hello world! 233+hello world, again 234------------------------------------------------ 235 236So "git diff" is comparing against something other than the head. 237The thing that it's comparing against is actually the index file, 238which is stored in .git/index in a binary format, but whose contents 239we can examine with ls-files: 240 241------------------------------------------------ 242$ git ls-files --stage 243100644 513feba2e53ebbd2532419ded848ba19de88ba00 0 file.txt 244$ git cat-file -t 513feba2 245blob 246$ git cat-file blob 513feba2 247hello world, again 248------------------------------------------------ 249 250So what our "git update-index" did was store a new blob and then put 251a reference to it in the index file. If we modify the file again, 252we'll see that the new modifications are reflected in the "git-diff" 253output: 254 255------------------------------------------------ 256$ echo 'again?' >>file.txt 257$ git diff 258index 513feba..ba3da7b 100644 259--- a/file.txt 260+++ b/file.txt 261@@ -1,2 +1,3 @@ 262 hello world! 263 hello world, again 264+again? 265------------------------------------------------ 266 267With the right arguments, git diff can also show us the difference 268between the working directory and the last commit, or between the 269index and the last commit: 270 271------------------------------------------------ 272$ git diff HEAD 273diff --git a/file.txt b/file.txt 274index a042389..ba3da7b 100644 275--- a/file.txt 276+++ b/file.txt 277@@ -1 +1,3 @@ 278 hello world! 279+hello world, again 280+again? 281$ git diff --cached 282diff --git a/file.txt b/file.txt 283index a042389..513feba 100644 284--- a/file.txt 285+++ b/file.txt 286@@ -1 +1,2 @@ 287 hello world! 288+hello world, again 289------------------------------------------------ 290 291At any time, we can create a new commit using "git commit" (without 292the -a option), and verify that the state committed only includes the 293changes stored in the index file, not the additional change that is 294still only in our working tree: 295 296------------------------------------------------ 297$ git commit -m "repeat" 298$ git diff HEAD 299diff --git a/file.txt b/file.txt 300index 513feba..ba3da7b 100644 301--- a/file.txt 302+++ b/file.txt 303@@ -1,2 +1,3 @@ 304 hello world! 305 hello world, again 306+again? 307------------------------------------------------ 308 309So by default "git commit" uses the index to create the commit, not 310the working tree; the -a option to commit tells it to first update 311the index with all changes in the working tree. 312 313Finally, it's worth looking at the effect of "git add" on the index 314file: 315 316------------------------------------------------ 317$ echo "goodbye, world" >closing.txt 318$ git add closing.txt 319------------------------------------------------ 320 321The effect of the "git add" was to add one entry to the index file: 322 323------------------------------------------------ 324$ git ls-files --stage 325100644 8b9743b20d4b15be3955fc8d5cd2b09cd2336138 0 closing.txt 326100644 513feba2e53ebbd2532419ded848ba19de88ba00 0 file.txt 327------------------------------------------------ 328 329And, as you can see with cat-file, this new entry refers to the 330current contents of the file: 331 332------------------------------------------------ 333$ git cat-file blob a6b11f7a 334goodbye, word 335------------------------------------------------ 336 337The "status" command is a useful way to get a quick summary of the 338situation: 339 340------------------------------------------------ 341$ git status 342# 343# Updated but not checked in: 344# (will commit) 345# 346# new file: closing.txt 347# 348# 349# Changed but not updated: 350# (use git-update-index to mark for commit) 351# 352# modified: file.txt 353# 354------------------------------------------------ 355 356Since the current state of closing.txt is cached in the index file, 357it is listed as "updated but not checked in". Since file.txt has 358changes in the working directory that aren't reflected in the index, 359it is marked "changed but not updated". At this point, running "git 360commit" would create a commit that added closing.txt (with its new 361contents), but that didn't modify file.txt. 362 363Also, note that a bare "git diff" shows the changes to file.txt, but 364not the addition of closing.txt, because the version of closing.txt 365in the index file is identical to the one in the working directory. 366 367In addition to being the staging area for new commits, the index file 368is also populated from the object database when checking out a 369branch, and is used to hold the trees involved in a merge operation. 370See the link:core-tutorial.txt[core tutorial] and the relevant man 371pages for details. 372 373What next? 374---------- 375 376At this point you should know everything necessary to read the man 377pages for any of the git commands; one good place to start would be 378with the commands mentioned in link:everyday.html[Everyday git]. You 379should be able to find any unknown jargon in the 380link:glossary.html[Glosssay]. 381 382The link:cvs-migration.html[CVS migration] document explains how to 383import a CVS repository into git, and shows how to use git in a 384CVS-like way. 385 386For some interesting examples of git use, see the 387link:howto-index.html[howtos]. 388 389For git developers, the link:core-tutorial.html[Core tutorial] goes 390into detail on the lower-level git mechanisms involved in, for 391example, creating a new commit.