4 This directory holds many test scripts for core GIT tools. The
5 first part of this short document describes how to run the tests
8 When fixing the tools or adding enhancements, you are strongly
9 encouraged to add tests in this directory to cover what you are
10 trying to fix or enhance. The later part of this short document
11 describes how your test scripts should be organized.
17 The easiest way to run tests is to say "make". This runs all
20 *** t0000-basic.sh ***
21 ok 1 - .git/objects should be empty after git init in an empty repo.
22 ok 2 - .git/objects should have 3 subdirectories.
23 ok 3 - success is reported like this
25 ok 43 - very long name in the index handled sanely
26 # fixed 1 known breakage(s)
27 # still have 1 known breakage(s)
28 # passed all remaining 42 test(s)
32 ok 2 - plain with GIT_WORK_TREE
35 Since the tests all output TAP (see http://testanything.org) they can
36 be run with any TAP harness. Here's an example of parallel testing
37 powered by a recent version of prove(1):
39 $ prove --timer --jobs 15 ./t[0-9]*.sh
40 [19:17:33] ./t0005-signals.sh ................................... ok 36 ms
41 [19:17:33] ./t0022-crlf-rename.sh ............................... ok 69 ms
42 [19:17:33] ./t0024-crlf-archive.sh .............................. ok 154 ms
43 [19:17:33] ./t0004-unwritable.sh ................................ ok 289 ms
44 [19:17:33] ./t0002-gitfile.sh ................................... ok 480 ms
45 ===( 102;0 25/? 6/? 5/? 16/? 1/? 4/? 2/? 1/? 3/? 1... )===
47 prove and other harnesses come with a lot of useful options. The
48 --state option in particular is very useful:
50 # Repeat until no more failures
51 $ prove -j 15 --state=failed,save ./t[0-9]*.sh
53 You can give DEFAULT_TEST_TARGET=prove on the make command (or define it
54 in config.mak) to cause "make test" to run tests under prove.
55 GIT_PROVE_OPTS can be used to pass additional options, e.g.
57 $ make DEFAULT_TEST_TARGET=prove GIT_PROVE_OPTS='--timer --jobs 16' test
59 You can also run each test individually from command line, like this:
61 $ sh ./t3010-ls-files-killed-modified.sh
62 ok 1 - git update-index --add to add various paths.
63 ok 2 - git ls-files -k to show killed files.
64 ok 3 - validate git ls-files -k output.
65 ok 4 - git ls-files -m to show modified files.
66 ok 5 - validate git ls-files -m output.
67 # passed all 5 test(s)
70 You can pass --verbose (or -v), --debug (or -d), and --immediate
71 (or -i) command line argument to the test, or by setting GIT_TEST_OPTS
72 appropriately before running "make".
75 This makes the test more verbose. Specifically, the
76 command being run and their output if any are also
79 --verbose-only=<pattern>::
80 Like --verbose, but the effect is limited to tests with
81 numbers matching <pattern>. The number matched against is
82 simply the running count of the test within the file.
85 This may help the person who is developing a new test.
86 It causes the command defined with test_debug to run.
87 The "trash" directory (used to store all temporary data
88 during testing) is not deleted even if there are no
89 failed tests so that you can inspect its contents after
93 This causes the test to immediately exit upon the first
94 failed test. Cleanup commands requested with
95 test_when_finished are not executed if the test failed,
96 in order to keep the state for inspection by the tester
100 This causes additional long-running tests to be run (where
101 available), for more exhaustive testing.
104 Execute all Git binaries under valgrind tool <tool> and exit
105 with status 126 on errors (just like regular tests, this will
106 only stop the test script when running under -i).
108 Since it makes no sense to run the tests with --valgrind and
109 not see any output, this option implies --verbose. For
110 convenience, it also implies --tee.
112 <tool> defaults to 'memcheck', just like valgrind itself.
113 Other particularly useful choices include 'helgrind' and
114 'drd', but you may use any tool recognized by your valgrind
117 As a special case, <tool> can be 'memcheck-fast', which uses
118 memcheck but disables --track-origins. Use this if you are
119 running tests in bulk, to see if there are _any_ memory
122 Note that memcheck is run with the option --leak-check=no,
123 as the git process is short-lived and some errors are not
124 interesting. In order to run a single command under the same
125 conditions manually, you should set GIT_VALGRIND to point to
126 the 't/valgrind/' directory and use the commands under
129 --valgrind-only=<pattern>::
130 Like --valgrind, but the effect is limited to tests with
131 numbers matching <pattern>. The number matched against is
132 simply the running count of the test within the file.
135 In addition to printing the test output to the terminal,
136 write it to files named 't/test-results/$TEST_NAME.out'.
137 As the names depend on the tests' file names, it is safe to
138 run the tests with this option in parallel.
141 By default tests are run without dashed forms of
142 commands (like git-commit) in the PATH (it only uses
143 wrappers from ../bin-wrappers). Use this option to include
144 the build directory (..) in the PATH, which contains all
145 the dashed forms of commands. This option is currently
146 implied by other options like --valgrind and
150 Create "trash" directories used to store all temporary data during
151 testing under <directory>, instead of the t/ directory.
152 Using this option with a RAM-based filesystem (such as tmpfs)
153 can massively speed up the test suite.
155 You can also set the GIT_TEST_INSTALLED environment variable to
156 the bindir of an existing git installation to test that installation.
157 You still need to have built this git sandbox, from which various
158 test-* support programs, templates, and perl libraries are used.
159 If your installed git is incomplete, it will silently test parts of
160 your built version instead.
162 When using GIT_TEST_INSTALLED, you can also set GIT_TEST_EXEC_PATH to
163 override the location of the dashed-form subcommands (what
164 GIT_EXEC_PATH would be used for during normal operation).
165 GIT_TEST_EXEC_PATH defaults to `$GIT_TEST_INSTALLED/git --exec-path`.
171 In some environments, certain tests have no way of succeeding
172 due to platform limitation, such as lack of 'unzip' program, or
173 filesystem that do not allow arbitrary sequence of non-NUL bytes
176 You should be able to say something like
178 $ GIT_SKIP_TESTS=t9200.8 sh ./t9200-git-cvsexport-commit.sh
182 $ GIT_SKIP_TESTS='t[0-4]??? t91?? t9200.8' make
184 to omit such tests. The value of the environment variable is a
185 SP separated list of patterns that tells which tests to skip,
186 and either can match the "t[0-9]{4}" part to skip the whole
187 test, or t[0-9]{4} followed by ".$number" to say which
188 particular test to skip.
190 Note that some tests in the existing test suite rely on previous
191 test item, so you cannot arbitrarily disable one and expect the
192 remainder of test to check what the test originally was intended
199 The test files are named as:
201 tNNNN-commandname-details.sh
203 where N is a decimal digit.
205 First digit tells the family:
207 0 - the absolute basics and global stuff
208 1 - the basic commands concerning database
209 2 - the basic commands concerning the working tree
210 3 - the other basic commands (e.g. ls-files)
211 4 - the diff commands
212 5 - the pull and exporting commands
213 6 - the revision tree commands (even e.g. merge-base)
214 7 - the porcelainish commands concerning the working tree
215 8 - the porcelainish commands concerning forensics
218 Second digit tells the particular command we are testing.
220 Third digit (optionally) tells the particular switch or group of switches
223 If you create files under t/ directory (i.e. here) that is not
224 the top-level test script, never name the file to match the above
225 pattern. The Makefile here considers all such files as the
226 top-level test script and tries to run all of them. Care is
227 especially needed if you are creating a common test library
228 file, similar to test-lib.sh, because such a library file may
229 not be suitable for standalone execution.
235 The test script is written as a shell script. It should start
236 with the standard "#!/bin/sh" with copyright notices, and an
237 assignment to variable 'test_description', like this:
241 # Copyright (c) 2005 Junio C Hamano
244 test_description='xxx test (option --frotz)
246 This test registers the following structure in the cache
247 and tries to run git-ls-files with option --frotz.'
253 After assigning test_description, the test script should source
254 test-lib.sh like this:
258 This test harness library does the following things:
260 - If the script is invoked with command line argument --help
261 (or -h), it shows the test_description and exits.
263 - Creates an empty test directory with an empty .git/objects database
264 and chdir(2) into it. This directory is 't/trash
265 directory.$test_name_without_dotsh', with t/ subject to change by
266 the --root option documented above.
268 - Defines standard test helper functions for your scripts to
269 use. These functions are designed to make all scripts behave
270 consistently when command line arguments --verbose (or -v),
271 --debug (or -d), and --immediate (or -i) is given.
273 Do's, don'ts & things to keep in mind
274 -------------------------------------
276 Here are a few examples of things you probably should and shouldn't do
281 - Put all code inside test_expect_success and other assertions.
283 Even code that isn't a test per se, but merely some setup code
284 should be inside a test assertion.
286 - Chain your test assertions
288 Write test code like this:
300 That way all of the commands in your tests will succeed or fail. If
301 you must ignore the return value of something, consider using a
302 helper function (e.g. use sane_unset instead of unset, in order
303 to avoid unportable return value for unsetting a variable that was
304 already unset), or prepending the command with test_might_fail or
307 - Check the test coverage for your tests. See the "Test coverage"
310 Don't blindly follow test coverage metrics; if a new function you added
311 doesn't have any coverage, then you're probably doing something wrong,
312 but having 100% coverage doesn't necessarily mean that you tested
315 Tests that are likely to smoke out future regressions are better
316 than tests that just inflate the coverage metrics.
318 - When a test checks for an absolute path that a git command generated,
319 construct the expected value using $(pwd) rather than $PWD,
320 $TEST_DIRECTORY, or $TRASH_DIRECTORY. It makes a difference on
321 Windows, where the shell (MSYS bash) mangles absolute path names.
322 For details, see the commit message of 4114156ae9.
326 - exit() within a <script> part.
328 The harness will catch this as a programming error of the test.
329 Use test_done instead if you need to stop the tests early (see
330 "Skipping tests" below).
332 - use '! git cmd' when you want to make sure the git command exits
333 with failure in a controlled way by calling "die()". Instead,
334 use 'test_must_fail git cmd'. This will signal a failure if git
335 dies in an unexpected way (e.g. segfault).
337 On the other hand, don't use test_must_fail for running regular
338 platform commands; just use '! cmd'.
340 - use perl without spelling it as "$PERL_PATH". This is to help our
341 friends on Windows where the platform Perl often adds CR before
342 the end of line, and they bundle Git with a version of Perl that
343 does not do so, whose path is specified with $PERL_PATH. Note that we
344 provide a "perl" function which uses $PERL_PATH under the hood, so
345 you do not need to worry when simply running perl in the test scripts
346 (but you do, for example, on a shebang line or in a sub script
347 created via "write_script").
349 - use sh without spelling it as "$SHELL_PATH", when the script can
350 be misinterpreted by broken platform shell (e.g. Solaris).
352 - chdir around in tests. It is not sufficient to chdir to
353 somewhere and then chdir back to the original location later in
354 the test, as any intermediate step can fail and abort the test,
355 causing the next test to start in an unexpected directory. Do so
356 inside a subshell if necessary.
358 - Break the TAP output
360 The raw output from your test may be interpreted by a TAP harness. TAP
361 harnesses will ignore everything they don't know about, but don't step
362 on their toes in these areas:
364 - Don't print lines like "$x..$y" where $x and $y are integers.
366 - Don't print lines that begin with "ok" or "not ok".
368 TAP harnesses expect a line that begins with either "ok" and "not
369 ok" to signal a test passed or failed (and our harness already
370 produces such lines), so your script shouldn't emit such lines to
373 You can glean some further possible issues from the TAP grammar
374 (see http://search.cpan.org/perldoc?TAP::Parser::Grammar#TAP_Grammar)
375 but the best indication is to just run the tests with prove(1),
376 it'll complain if anything is amiss.
380 - Inside <script> part, the standard output and standard error
381 streams are discarded, and the test harness only reports "ok" or
382 "not ok" to the end user running the tests. Under --verbose, they
383 are shown to help debugging the tests.
389 If you need to skip tests you should do so by using the three-arg form
390 of the test_* functions (see the "Test harness library" section
393 test_expect_success PERL 'I need Perl' '
394 perl -e "hlagh() if unf_unf()"
397 The advantage of skipping tests like this is that platforms that don't
398 have the PERL and other optional dependencies get an indication of how
399 many tests they're missing.
401 If the test code is too hairy for that (i.e. does a lot of setup work
402 outside test assertions) you can also skip all remaining tests by
403 setting skip_all and immediately call test_done:
405 if ! test_have_prereq PERL
407 skip_all='skipping perl interface tests, perl not available'
411 The string you give to skip_all will be used as an explanation for why
412 the test was skipped.
417 Your script will be a sequence of tests, using helper functions
418 from the test harness library. At the end of the script, call
425 There are a handful helper functions defined in the test harness
426 library for your script to use.
428 - test_expect_success [<prereq>] <message> <script>
430 Usually takes two strings as parameters, and evaluates the
431 <script>. If it yields success, test is considered
432 successful. <message> should state what it is testing.
436 test_expect_success \
437 'git-write-tree should be able to write an empty tree.' \
438 'tree=$(git-write-tree)'
440 If you supply three parameters the first will be taken to be a
441 prerequisite; see the test_set_prereq and test_have_prereq
444 test_expect_success TTY 'git --paginate rev-list uses a pager' \
447 You can also supply a comma-separated list of prerequisites, in the
448 rare case where your test depends on more than one:
450 test_expect_success PERL,PYTHON 'yo dawg' \
451 ' test $(perl -E 'print eval "1 +" . qx[python -c "print 2"]') == "4" '
453 - test_expect_failure [<prereq>] <message> <script>
455 This is NOT the opposite of test_expect_success, but is used
456 to mark a test that demonstrates a known breakage. Unlike
457 the usual test_expect_success tests, which say "ok" on
458 success and "FAIL" on failure, this will say "FIXED" on
459 success and "still broken" on failure. Failures from these
460 tests won't cause -i (immediate) to stop.
462 Like test_expect_success this function can optionally use a three
463 argument invocation with a prerequisite as the first argument.
465 - test_debug <script>
467 This takes a single argument, <script>, and evaluates it only
468 when the test script is started with --debug command line
469 argument. This is primarily meant for use during the
470 development of a new test script.
474 Your test script must have test_done at the end. Its purpose
475 is to summarize successes and failures in the test script and
476 exit with an appropriate error code.
480 Make commit and tag names consistent by setting the author and
481 committer times to defined state. Subsequent calls will
482 advance the times by a fixed amount.
484 - test_commit <message> [<filename> [<contents>]]
486 Creates a commit with the given message, committing the given
487 file with the given contents (default for both is to reuse the
488 message string), and adds a tag (again reusing the message
489 string as name). Calls test_tick to make the SHA-1s
492 - test_merge <message> <commit-or-tag>
494 Merges the given rev using the given message. Like test_commit,
495 creates a tag and calls test_tick before committing.
497 - test_set_prereq <prereq>
499 Set a test prerequisite to be used later with test_have_prereq. The
500 test-lib will set some prerequisites for you, see the
501 "Prerequisites" section below for a full list of these.
503 Others you can set yourself and use later with either
504 test_have_prereq directly, or the three argument invocation of
505 test_expect_success and test_expect_failure.
507 - test_have_prereq <prereq>
509 Check if we have a prerequisite previously set with
510 test_set_prereq. The most common use of this directly is to skip
511 all the tests if we don't have some essential prerequisite:
513 if ! test_have_prereq PERL
515 skip_all='skipping perl interface tests, perl not available'
519 - test_external [<prereq>] <message> <external> <script>
521 Execute a <script> with an <external> interpreter (like perl). This
522 was added for tests like t9700-perl-git.sh which do most of their
523 work in an external test script.
526 'GitwebCache::*FileCache*' \
527 perl "$TEST_DIRECTORY"/t9503/test_cache_interface.pl
529 If the test is outputting its own TAP you should set the
530 test_external_has_tap variable somewhere before calling the first
531 test_external* function. See t9700-perl-git.sh for an example.
533 # The external test will outputs its own plan
534 test_external_has_tap=1
536 - test_external_without_stderr [<prereq>] <message> <external> <script>
538 Like test_external but fail if there's any output on stderr,
539 instead of checking the exit code.
541 test_external_without_stderr \
543 perl "$TEST_DIRECTORY"/t9700/test.pl
545 - test_expect_code <exit-code> <command>
547 Run a command and ensure that it exits with the given exit code.
550 test_expect_success 'Merge with d/f conflicts' '
551 test_expect_code 1 git merge "merge msg" B master
554 - test_must_fail <git-command>
556 Run a git command and ensure it fails in a controlled way. Use
557 this instead of "! <git-command>". When git-command dies due to a
558 segfault, test_must_fail diagnoses it as an error; "! <git-command>"
559 treats it as just another expected failure, which would let such a
562 - test_might_fail <git-command>
564 Similar to test_must_fail, but tolerate success, too. Use this
565 instead of "<git-command> || :" to catch failures due to segv.
567 - test_cmp <expected> <actual>
569 Check whether the content of the <actual> file matches the
570 <expected> file. This behaves like "cmp" but produces more
571 helpful output when the test is run with "-v" option.
573 - test_line_count (= | -lt | -ge | ...) <length> <file>
575 Check whether a file has the length it is expected to.
577 - test_path_is_file <path> [<diagnosis>]
578 test_path_is_dir <path> [<diagnosis>]
579 test_path_is_missing <path> [<diagnosis>]
581 Check if the named path is a file, if the named path is a
582 directory, or if the named path does not exist, respectively,
583 and fail otherwise, showing the <diagnosis> text.
585 - test_when_finished <script>
587 Prepend <script> to a list of commands to run to clean up
588 at the end of the current test. If some clean-up command
589 fails, the test will not pass.
593 test_expect_success 'branch pointing to non-commit' '
594 git rev-parse HEAD^{tree} >.git/refs/heads/invalid &&
595 test_when_finished "git update-ref -d refs/heads/invalid" &&
601 This command is useful for writing and debugging tests and must be
602 removed before submitting. It halts the execution of the test and
603 spawns a shell in the trash directory. Exit the shell to continue
606 test_expect_success 'test' '
607 git do-something >actual &&
609 test_cmp expected actual
612 - test_ln_s_add <path1> <path2>
614 This function helps systems whose filesystem does not support symbolic
615 links. Use it to add a symbolic link entry to the index when it is not
616 important that the file system entry is a symbolic link, i.e., instead
622 Sometimes it is possible to split a test in a part that does not need
623 the symbolic link in the file system and a part that does; then only
624 the latter part need be protected by a SYMLINKS prerequisite (see below).
629 These are the prerequisites that the test library predefines with
632 See the prereq argument to the test_* functions in the "Test harness
633 library" section above and the "test_have_prereq" function for how to
634 use these, and "test_set_prereq" for how to define your own.
638 Git wasn't compiled with NO_PYTHON=YesPlease. Wrap any tests that
639 need Python with this.
643 Git wasn't compiled with NO_PERL=YesPlease.
645 Even without the PERL prerequisite, tests can assume there is a
646 usable perl interpreter at $PERL_PATH, though it need not be
651 The filesystem supports POSIX style permission bits.
655 Backslashes in pathspec are not directory separators. This is not
656 set on Windows. See 6fd1106a for details.
660 The process retains the same pid across exec(2). See fb9a2bea for
665 The filesystem we're on supports creation of FIFOs (named pipes)
670 The filesystem we're on supports symbolic links. E.g. a FAT
671 filesystem doesn't support these. See 704a3143 for details.
675 Test is not run by root user, and an attempt to write to an
676 unwritable file is expected to fail correctly.
680 Git was compiled with USE_LIBPCRE=YesPlease. Wrap any tests
681 that use git-grep --perl-regexp or git-grep -P in these.
683 - CASE_INSENSITIVE_FS
685 Test is run on a case insensitive file system.
689 Test is run on a filesystem which converts decomposed utf-8 (nfd)
690 to precomposed utf-8 (nfc).
692 Tips for Writing Tests
693 ----------------------
695 As with any programming projects, existing programs are the best
696 source of the information. However, do _not_ emulate
697 t0000-basic.sh when writing your tests. The test is special in
698 that it tries to validate the very core of GIT. For example, it
699 knows that there will be 256 subdirectories under .git/objects/,
700 and it knows that the object ID of an empty tree is a certain
701 40-byte string. This is deliberately done so in t0000-basic.sh
702 because the things the very basic core test tries to achieve is
703 to serve as a basis for people who are changing the GIT internal
704 drastically. For these people, after making certain changes,
705 not seeing failures from the basic test _is_ a failure. And
706 such drastic changes to the core GIT that even changes these
707 otherwise supposedly stable object IDs should be accompanied by
708 an update to t0000-basic.sh.
710 However, other tests that simply rely on basic parts of the core
711 GIT working properly should not have that level of intimate
712 knowledge of the core GIT internals. If all the test scripts
713 hardcoded the object IDs like t0000-basic.sh does, that defeats
714 the purpose of t0000-basic.sh, which is to isolate that level of
715 validation in one place. Your test also ends up needing
716 updating when such a change to the internal happens, so do _not_
717 do it and leave the low level of validation to t0000-basic.sh.
722 You can use the coverage tests to find code paths that are not being
723 used or properly exercised yet.
725 To do that, run the coverage target at the top-level (not in the t/
730 That'll compile Git with GCC's coverage arguments, and generate a test
731 report with gcov after the tests finish. Running the coverage tests
732 can take a while, since running the tests in parallel is incompatible
733 with GCC's coverage mode.
735 After the tests have run you can generate a list of untested
738 make coverage-untested-functions
740 You can also generate a detailed per-file HTML report using the
741 Devel::Cover module. To install it do:
743 # On Debian or Ubuntu:
744 sudo aptitude install libdevel-cover-perl
746 # From the CPAN with cpanminus
747 curl -L http://cpanmin.us | perl - --sudo --self-upgrade
748 cpanm --sudo Devel::Cover
750 Then, at the top-level:
754 That'll generate a detailed cover report in the "cover_db_html"
755 directory, which you can then copy to a webserver, or inspect locally