tutorial: expanded discussion of commit history
Expand the history-browsing section of the tutorial a bit, in part to address Junio's suggestion that we mention "git grep" and Linus's complaint that people are missing the flexibility of the commandline interfaces for selecting commits. This reads a little more like a collection of examples than a "tutorial", but maybe that's what people need at this point. Signed-off-by: J. Bruce Fields <bfields@citi.umich.edu> Signed-off-by: Junio C Hamano <junkio@cox.net>
J. Bruce Fields committed
May 21, 2006 at 16:54 UTC
f1fe3846e4236a05879d7f84b11ae8ea64dd8d0d
1 file changed
+112
-53
Documentation/tutorial.txt
+112
-53
@@ -288,103 +288,162 @@ Git can also be used in a CVS-like mode, with a central repository
288
that various users push changes to; see gitlink:git-push[1] and
289
link:cvs-migration.html[git for CVS users].
290
291
-Keeping track of history
292
-------------------------
291
+Exploring history
292
+-----------------
293
294
-Git history is represented as a series of interrelated commits. The
295
-most recent commit in the currently checked-out branch can always be
296
-referred to as HEAD, and the "parent" of any commit can always be
297
-referred to by appending a caret, "^", to the end of the name of the
298
-commit. So, for example,
294
+Git history is represented as a series of interrelated commits. We
295
+have already seen that the git log command can list those commits.
296
+Note that first line of each git log entry also gives a name for the
297
+commit:
298
299
-------------------------------------
301
-git diff HEAD^ HEAD
300
+$ git log
301
+commit c82a22c39cbc32576f64f5c6b3f24b99ea8149c7
302
+Author: Junio C Hamano <junkio@cox.net>
303
+Date: Tue May 16 17:18:22 2006 -0700
304
+
305
+ merge-base: Clarify the comments on post processing.
306
-------------------------------------
307
304
-shows the difference between the most-recently checked-in state of
305
-the tree and the previous state, and
308
+We can give this name to git show to see the details about this
309
+commit.
310
311
-------------------------------------
308
-git diff HEAD^^ HEAD^
312
+$ git show c82a22c39cbc32576f64f5c6b3f24b99ea8149c7
313
-------------------------------------
314
311
-shows the difference between that previous state and the state two
312
-commits ago. Also, HEAD~5 can be used as a shorthand for HEAD{caret}{caret}{caret}{caret}{caret},
313
-and more generally HEAD~n can refer to the nth previous commit.
314
-Commits representing merges have more than one parent, and you can
315
-specify which parent to follow in that case; see
316
-gitlink:git-rev-parse[1].
315
+But there other ways to refer to commits. You can use any initial
316
+part of the name that is long enough to uniquely identify the commit:
317
318
-The name of a branch can also be used to refer to the most recent
319
-commit on that branch; so you can also say things like
318
+-------------------------------------
319
+$ git show c82a22c39c # the first few characters of the name are
320
+ # usually enough
321
+$ git show HEAD # the tip of the current branch
322
+$ git show experimental # the tip of the "experimental" branch
323
+-------------------------------------
324
+
325
+Every commit has at least one "parent" commit, which points to the
326
+previous state of the project:
327
328
-------------------------------------
322
-git diff HEAD experimental
329
+$ git show HEAD^ # to see the parent of HEAD
330
+$ git show HEAD^^ # to see the grandparent of HEAD
331
+$ git show HEAD~4 # to see the great-great grandparent of HEAD
332
-------------------------------------
333
325
-to see the difference between the most-recently committed tree in
326
-the current branch and the most-recently committed tree in the
327
-experimental branch.
334
+Note that merge commits may have more than one parent:
335
+
336
+-------------------------------------
337
+$ git show HEAD^1 # show the first parent of HEAD (same as HEAD^)
338
+$ git show HEAD^2 # show the second parent of HEAD
339
+-------------------------------------
340
329
-But you may find it more useful to see the list of commits made in
330
-the experimental branch but not in the current branch, and
341
+You can also give commits names of your own; after running
342
343
-------------------------------------
333
-git log HEAD..experimental
344
+$ git-tag v2.5 1b2e1d63ff
345
-------------------------------------
346
336
-will do that, just as
347
+you can refer to 1b2e1d63ff by the name "v2.5". If you intend to
348
+share this name with other people (for example, to identify a release
349
+version), you should create a "tag" object, and perhaps sign it; see
350
+gitlink:git-tag[1] for details.
351
+
352
+Any git command that needs to know a commit can take any of these
353
+names. For example:
354
355
-------------------------------------
339
-git log experimental..HEAD
356
+$ git diff v2.5 HEAD # compare the current HEAD to v2.5
357
+$ git branch stable v2.5 # start a new branch named "stable" based
358
+ # at v2.5
359
+$ git reset --hard HEAD^ # reset your current branch and working
360
+ # directory its state at HEAD^
361
-------------------------------------
362
342
-will show the list of commits made on the HEAD but not included in
343
-experimental.
363
+Be careful with that last command: in addition to losing any changes
364
+in the working directory, it will also remove all later commits from
365
+this branch. If this branch is the only branch containing those
366
+commits, they will be lost. (Also, don't use "git reset" on a
367
+publicly-visible branch that other developers pull from, as git will
368
+be confused by history that disappears in this way.)
369
345
-You can also give commits convenient names of your own: after running
370
+The git grep command can search for strings in any version of your
371
+project, so
372
373
-------------------------------------
348
-$ git-tag v2.5 HEAD^^
374
+$ git grep "hello" v2.5
375
-------------------------------------
376
351
-you can refer to HEAD^^ by the name "v2.5". If you intend to share
352
-this name with other people (for example, to identify a release
353
-version), you should create a "tag" object, and perhaps sign it; see
354
-gitlink:git-tag[1] for details.
377
+searches for all occurences of "hello" in v2.5.
378
356
-You can revisit the old state of a tree, and make further
357
-modifications if you wish, using git branch: the command
379
+If you leave out the commit name, git grep will search any of the
380
+files it manages in your current directory. So
381
382
-------------------------------------
360
-$ git branch stable-release v2.5
383
+$ git grep "hello"
384
-------------------------------------
385
363
-will create a new branch named "stable-release" starting from the
364
-commit which you tagged with the name v2.5.
386
+is a quick way to search just the files that are tracked by git.
387
366
-You can reset the state of any branch to an earlier commit at any
367
-time with
388
+Many git commands also take sets of commits, which can be specified
389
+in a number of ways. Here are some examples with git log:
390
391
-------------------------------------
370
-$ git reset --hard v2.5
392
+$ git log v2.5..v2.6 # commits between v2.5 and v2.6
393
+$ git log v2.5.. # commits since v2.5
394
+$ git log --since="2 weeks ago" # commits from the last 2 weeks
395
+$ git log v2.5.. Makefile # commits since v2.5 which modify
396
+ # Makefile
397
-------------------------------------
398
373
-This will remove all later commits from this branch and reset the
374
-working tree to the state it had when the given commit was made. If
375
-this branch is the only branch containing the later commits, those
376
-later changes will be lost. Don't use "git reset" on a
377
-publicly-visible branch that other developers pull from, as git will
378
-be confused by history that disappears in this way.
399
+You can also give git log a "range" of commits where the first is not
400
+necessarily an ancestor of the second; for example, if the tips of
401
+the branches "stable-release" and "master" diverged from a common
402
+commit some time ago, then
403
+
404
+-------------------------------------
405
+$ git log stable..experimental
406
+-------------------------------------
407
+
408
+will list commits made in the experimental branch but not in the
409
+stable branch, while
410
+
411
+-------------------------------------
412
+$ git log experimental..stable
413
+-------------------------------------
414
+
415
+will show the list of commits made on the stable branch but not
416
+the experimental branch.
417
+
418
+The "git log" command has a weakness: it must present commits in a
419
+list. When the history has lines of development that diverged and
420
+then merged back together, the order in which "git log" presents
421
+those commits is meaningless.
422
+
423
+Most projects with multiple contributors (such as the linux kernel,
424
+or git itself) have frequent merges, and gitk does a better job of
425
+visualizing their history. For example,
426
+
427
+-------------------------------------
428
+$ gitk --since="2 weeks ago" drivers/
429
+-------------------------------------
430
+
431
+allows you to browse any commits from the last 2 weeks of commits
432
+that modified files under the "drivers" directory.
433
+
434
+Finally, most commands that take filenames will optionally allow you
435
+to precede any filename by a commit, to specify a particular version
436
+fo the file:
437
+
438
+-------------------------------------
439
+$ git diff v2.5:Makefile HEAD:Makefile.in
440
+-------------------------------------
441
442
Next Steps
443
----------
444
445
Some good commands to explore next:
446
385
- * gitlink:git-diff[1]: This flexible command does much more than
386
- we've seen in the few examples above.
387
-
447
* gitlink:git-format-patch[1], gitlink:git-am[1]: These convert
448
series of git commits into emailed patches, and vice versa,
449
useful for projects such as the linux kernel which rely heavily