blob: 1a2072b2c8a282c48f5b15700428ebed2f2dfea1 [file] [log] [blame]
Denton Liu788db142020-06-07 05:48:26 -04001Core Git Tests
Junio C Hamano986aa7f2005-05-14 00:25:06 -07002==============
3
Denton Liu788db142020-06-07 05:48:26 -04004This directory holds many test scripts for core Git tools. The
Junio C Hamano986aa7f2005-05-14 00:25:06 -07005first part of this short document describes how to run the tests
6and read their output.
7
8When fixing the tools or adding enhancements, you are strongly
9encouraged to add tests in this directory to cover what you are
10trying to fix or enhance. The later part of this short document
11describes how your test scripts should be organized.
12
13
14Running Tests
15-------------
16
17The easiest way to run tests is to say "make". This runs all
18the tests.
19
20 *** t0000-basic.sh ***
Ævar Arnfjörð Bjarmason5099b992010-06-24 21:52:12 +000021 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
Junio C Hamano986aa7f2005-05-14 00:25:06 -070024 ...
Ævar Arnfjörð Bjarmason5099b992010-06-24 21:52:12 +000025 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)
29 1..43
30 *** t0001-init.sh ***
31 ok 1 - plain
32 ok 2 - plain with GIT_WORK_TREE
33 ok 3 - plain bare
Junio C Hamano986aa7f2005-05-14 00:25:06 -070034
Ævar Arnfjörð Bjarmason5099b992010-06-24 21:52:12 +000035Since the tests all output TAP (see http://testanything.org) they can
Ævar Arnfjörð Bjarmason85b0b342010-07-02 14:59:44 +000036be run with any TAP harness. Here's an example of parallel testing
Ævar Arnfjörð Bjarmason5099b992010-06-24 21:52:12 +000037powered by a recent version of prove(1):
Junio C Hamano986aa7f2005-05-14 00:25:06 -070038
Ævar Arnfjörð Bjarmason5099b992010-06-24 21:52:12 +000039 $ 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... )===
46
47prove and other harnesses come with a lot of useful options. The
48--state option in particular is very useful:
49
50 # Repeat until no more failures
51 $ prove -j 15 --state=failed,save ./t[0-9]*.sh
52
Michael J Gruber28d836c2010-10-14 10:53:36 +020053You can give DEFAULT_TEST_TARGET=prove on the make command (or define it
54in config.mak) to cause "make test" to run tests under prove.
55GIT_PROVE_OPTS can be used to pass additional options, e.g.
56
57 $ make DEFAULT_TEST_TARGET=prove GIT_PROVE_OPTS='--timer --jobs 16' test
58
Ævar Arnfjörð Bjarmason5099b992010-06-24 21:52:12 +000059You can also run each test individually from command line, like this:
60
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)
68 1..5
Junio C Hamano986aa7f2005-05-14 00:25:06 -070069
70You can pass --verbose (or -v), --debug (or -d), and --immediate
Johannes Schindelin4e1be632009-02-04 00:25:59 +010071(or -i) command line argument to the test, or by setting GIT_TEST_OPTS
Matheus Tavares78dc0882020-03-22 12:58:57 -030072appropriately before running "make". Short options can be bundled, i.e.
73'-d -v' is the same as '-dv'.
Junio C Hamano986aa7f2005-05-14 00:25:06 -070074
Ilya Bobyr5e3b4fc2014-04-30 02:50:42 -070075-v::
Junio C Hamano986aa7f2005-05-14 00:25:06 -070076--verbose::
77 This makes the test more verbose. Specifically, the
78 command being run and their output if any are also
79 output.
80
Thomas Rastff09af32013-06-23 20:12:56 +020081--verbose-only=<pattern>::
82 Like --verbose, but the effect is limited to tests with
83 numbers matching <pattern>. The number matched against is
84 simply the running count of the test within the file.
85
Jeff Kinga136f6d2014-10-10 02:47:27 -040086-x::
87 Turn on shell tracing (i.e., `set -x`) during the tests
SZEDER Gábor94201a22018-02-24 00:39:50 +010088 themselves. Implies `--verbose`.
SZEDER Gábor5fc98e72018-02-24 00:39:42 +010089 Ignored in test scripts that set the variable 'test_untraceable'
90 to a non-empty value, unless it's run with a Bash version
91 supporting BASH_XTRACEFD, i.e. v4.1 or later.
Jeff Kinga136f6d2014-10-10 02:47:27 -040092
Ilya Bobyr5e3b4fc2014-04-30 02:50:42 -070093-d::
Junio C Hamano986aa7f2005-05-14 00:25:06 -070094--debug::
95 This may help the person who is developing a new test.
96 It causes the command defined with test_debug to run.
Piotr Krukowiecki0986de92011-03-15 20:58:14 +010097 The "trash" directory (used to store all temporary data
98 during testing) is not deleted even if there are no
99 failed tests so that you can inspect its contents after
100 the test finished.
Junio C Hamano986aa7f2005-05-14 00:25:06 -0700101
Ilya Bobyr5e3b4fc2014-04-30 02:50:42 -0700102-i::
Junio C Hamano986aa7f2005-05-14 00:25:06 -0700103--immediate::
104 This causes the test to immediately exit upon the first
Simon Ruderich13cb3bb2013-04-09 23:48:36 +0200105 failed test. Cleanup commands requested with
106 test_when_finished are not executed if the test failed,
107 in order to keep the state for inspection by the tester
108 to diagnose the bug.
Junio C Hamano986aa7f2005-05-14 00:25:06 -0700109
Ilya Bobyr5e3b4fc2014-04-30 02:50:42 -0700110-l::
Lea Wiemann5e2c08c2008-06-17 03:29:02 +0200111--long-tests::
112 This causes additional long-running tests to be run (where
113 available), for more exhaustive testing.
114
Ilya Bobyr0445e6f2014-04-30 02:50:44 -0700115-r::
116--run=<test-selector>::
117 Run only the subset of tests indicated by
118 <test-selector>. See section "Skipping Tests" below for
119 <test-selector> syntax.
120
Thomas Rast952af352013-03-31 10:00:16 +0200121--valgrind=<tool>::
122 Execute all Git binaries under valgrind tool <tool> and exit
123 with status 126 on errors (just like regular tests, this will
124 only stop the test script when running under -i).
Junio C Hamano986aa7f2005-05-14 00:25:06 -0700125
Johannes Schindelin3da93652009-02-04 00:26:26 +0100126 Since it makes no sense to run the tests with --valgrind and
127 not see any output, this option implies --verbose. For
128 convenience, it also implies --tee.
129
Thomas Rast952af352013-03-31 10:00:16 +0200130 <tool> defaults to 'memcheck', just like valgrind itself.
131 Other particularly useful choices include 'helgrind' and
132 'drd', but you may use any tool recognized by your valgrind
133 installation.
134
Thomas Rast95d9d5e2013-03-31 10:00:17 +0200135 As a special case, <tool> can be 'memcheck-fast', which uses
136 memcheck but disables --track-origins. Use this if you are
137 running tests in bulk, to see if there are _any_ memory
138 issues.
139
Thomas Rast952af352013-03-31 10:00:16 +0200140 Note that memcheck is run with the option --leak-check=no,
Carlos Martín Nieto9aec68d2011-03-15 10:32:11 +0100141 as the git process is short-lived and some errors are not
142 interesting. In order to run a single command under the same
143 conditions manually, you should set GIT_VALGRIND to point to
144 the 't/valgrind/' directory and use the commands under
145 't/valgrind/bin/'.
146
Thomas Rast5dfc3682013-06-23 20:12:57 +0200147--valgrind-only=<pattern>::
148 Like --valgrind, but the effect is limited to tests with
149 numbers matching <pattern>. The number matched against is
150 simply the running count of the test within the file.
151
Johannes Schindelin44138552009-02-04 00:26:12 +0100152--tee::
153 In addition to printing the test output to the terminal,
154 write it to files named 't/test-results/$TEST_NAME.out'.
155 As the names depend on the tests' file names, it is safe to
156 run the tests with this option in parallel.
157
SZEDER Gábora5f52c62018-10-29 13:13:59 +0100158-V::
Jeff King452320f2016-10-21 06:48:00 -0400159--verbose-log::
160 Write verbose output to the same logfile as `--tee`, but do
161 _not_ write it to stdout. Unlike `--tee --verbose`, this option
162 is safe to use when stdout is being consumed by a TAP parser
163 like `prove`. Implies `--tee` and `--verbose`.
164
Matthew Ogilviee4597aa2009-12-02 22:14:06 -0700165--with-dashes::
166 By default tests are run without dashed forms of
167 commands (like git-commit) in the PATH (it only uses
168 wrappers from ../bin-wrappers). Use this option to include
169 the build directory (..) in the PATH, which contains all
170 the dashed forms of commands. This option is currently
171 implied by other options like --valgrind and
172 GIT_TEST_INSTALLED.
173
Johannes Schindelindd167a32019-01-29 06:19:37 -0800174--no-bin-wrappers::
175 By default, the test suite uses the wrappers in
176 `../bin-wrappers/` to execute `git` and friends. With this option,
177 `../git` and friends are run directly. This is not recommended
178 in general, as the wrappers contain safeguards to ensure that no
179 files from an installed Git are used, but can speed up test runs
180 especially on platforms where running shell scripts is expensive
181 (most notably, Windows).
182
Thomas Rast0d4dbcd2010-06-10 20:24:46 +0200183--root=<directory>::
184 Create "trash" directories used to store all temporary data during
185 testing under <directory>, instead of the t/ directory.
186 Using this option with a RAM-based filesystem (such as tmpfs)
187 can massively speed up the test suite.
188
Jeff Kingbb79af92015-03-20 06:05:48 -0400189--chain-lint::
190--no-chain-lint::
191 If --chain-lint is enabled, the test harness will check each
192 test to make sure that it properly "&&-chains" all commands (so
193 that a failure in the middle does not go unnoticed by the final
194 exit code of the test). This check is performed in addition to
195 running the tests themselves. You may also enable or disable
196 this feature by setting the GIT_TEST_CHAIN_LINT environment
197 variable to "1" or "0", respectively.
198
Jeff King2d86a962021-05-13 02:25:53 -0400199 A few test scripts disable some of the more advanced
200 chain-linting detection in the name of efficiency. You can
201 override this by setting the GIT_TEST_CHAIN_LINT_HARDER
202 environment variable to "1".
203
SZEDER Gáborfb7d1e32019-01-05 02:08:59 +0100204--stress::
SZEDER Gáborfb7d1e32019-01-05 02:08:59 +0100205 Run the test script repeatedly in multiple parallel jobs until
206 one of them fails. Useful for reproducing rare failures in
207 flaky tests. The number of parallel jobs is, in order of
Johannes Schindelinf5457372019-03-03 06:44:55 -0800208 precedence: the value of the GIT_TEST_STRESS_LOAD
SZEDER Gáborfb7d1e32019-01-05 02:08:59 +0100209 environment variable, or twice the number of available
210 processors (as shown by the 'getconf' utility), or 8.
211 Implies `--verbose -x --immediate` to get the most information
212 about the failure. Note that the verbose output of each test
213 job is saved to 't/test-results/$TEST_NAME.stress-<nr>.out',
214 and only the output of the failed test job is shown on the
215 terminal. The names of the trash directories get a
216 '.stress-<nr>' suffix, and the trash directory of the failed
217 test job is renamed to end with a '.stress-failed' suffix.
218
Johannes Schindelinf5457372019-03-03 06:44:55 -0800219--stress-jobs=<N>::
220 Override the number of parallel jobs. Implies `--stress`.
221
SZEDER Gábor76e27fb2019-02-08 12:50:45 +0100222--stress-limit=<N>::
223 When combined with --stress run the test script repeatedly
224 this many times in each of the parallel jobs or until one of
Johannes Schindelinde69e6f2019-03-03 06:44:54 -0800225 them fails, whichever comes first. Implies `--stress`.
SZEDER Gábor76e27fb2019-02-08 12:50:45 +0100226
Matthew Ogilviee160da72009-11-29 23:19:28 -0700227You can also set the GIT_TEST_INSTALLED environment variable to
228the bindir of an existing git installation to test that installation.
229You still need to have built this git sandbox, from which various
230test-* support programs, templates, and perl libraries are used.
231If your installed git is incomplete, it will silently test parts of
232your built version instead.
233
234When using GIT_TEST_INSTALLED, you can also set GIT_TEST_EXEC_PATH to
235override the location of the dashed-form subcommands (what
236GIT_EXEC_PATH would be used for during normal operation).
237GIT_TEST_EXEC_PATH defaults to `$GIT_TEST_INSTALLED/git --exec-path`.
238
239
Jakub Narebskifbd458a2008-06-20 23:10:50 +0200240Skipping Tests
241--------------
242
243In some environments, certain tests have no way of succeeding
244due to platform limitation, such as lack of 'unzip' program, or
245filesystem that do not allow arbitrary sequence of non-NUL bytes
246as pathnames.
247
248You should be able to say something like
249
250 $ GIT_SKIP_TESTS=t9200.8 sh ./t9200-git-cvsexport-commit.sh
251
252and even:
253
254 $ GIT_SKIP_TESTS='t[0-4]??? t91?? t9200.8' make
255
256to omit such tests. The value of the environment variable is a
257SP separated list of patterns that tells which tests to skip,
258and either can match the "t[0-9]{4}" part to skip the whole
259test, or t[0-9]{4} followed by ".$number" to say which
260particular test to skip.
261
Ilya Bobyr0445e6f2014-04-30 02:50:44 -0700262For an individual test suite --run could be used to specify that
263only some tests should be run or that some tests should be
264excluded from a run.
265
Elijah Newrenf21ac362020-10-18 00:23:45 +0000266The argument for --run, <test-selector>, is a list of description
267substrings or globs or individual test numbers or ranges with an
268optional negation prefix (of '!') that define what tests in a test
269suite to include (or exclude, if negated) in the run. A range is two
270numbers separated with a dash and matches a range of tests with both
271ends been included. You may omit the first or the second number to
272mean "from the first test" or "up to the very last test" respectively.
Ilya Bobyr0445e6f2014-04-30 02:50:44 -0700273
Elijah Newrenf21ac362020-10-18 00:23:45 +0000274The argument to --run is split on commas into separate strings,
275numbers, and ranges, and picks all tests that match any of the
276individual selection criteria. If the substring of the description
277text that you want to match includes a comma, use the glob character
278'?' instead. For example --run='rebase,merge?cherry-pick' would match
279on all tests that match either the glob *rebase* or the glob
280*merge?cherry-pick*.
Ilya Bobyr0445e6f2014-04-30 02:50:44 -0700281
282If --run starts with an unprefixed number or range the initial
283set of tests to run is empty. If the first item starts with '!'
284all the tests are added to the initial set. After initial set is
285determined every test number or range is added or excluded from
286the set one by one, from left to right.
287
Ilya Bobyr0445e6f2014-04-30 02:50:44 -0700288For example, to run only tests up to a specific test (21), one
289could do this:
290
291 $ sh ./t9200-git-cvsexport-commit.sh --run='1-21'
292
293or this:
294
295 $ sh ./t9200-git-cvsexport-commit.sh --run='-21'
296
297Common case is to run several setup tests (1, 2, 3) and then a
298specific test (21) that relies on that setup:
299
Elijah Newrenf21ac362020-10-18 00:23:45 +0000300 $ sh ./t9200-git-cvsexport-commit.sh --run='1,2,3,21'
Ilya Bobyr0445e6f2014-04-30 02:50:44 -0700301
302or:
303
304 $ sh ./t9200-git-cvsexport-commit.sh --run=1,2,3,21
305
306or:
307
Elijah Newrenf21ac362020-10-18 00:23:45 +0000308 $ sh ./t9200-git-cvsexport-commit.sh --run='-3,21'
Ilya Bobyr0445e6f2014-04-30 02:50:44 -0700309
Kaartic Sivaraam01e4be62017-09-17 10:18:15 +0000310As noted above, the test set is built by going through the items
311from left to right, so this:
Ilya Bobyr0445e6f2014-04-30 02:50:44 -0700312
Elijah Newrenf21ac362020-10-18 00:23:45 +0000313 $ sh ./t9200-git-cvsexport-commit.sh --run='1-4,!3'
Ilya Bobyr0445e6f2014-04-30 02:50:44 -0700314
Kaartic Sivaraam01e4be62017-09-17 10:18:15 +0000315will run tests 1, 2, and 4. Items that come later have higher
Ville Skyttä2e3a16b2016-08-09 11:53:38 +0300316precedence. It means that this:
Ilya Bobyr0445e6f2014-04-30 02:50:44 -0700317
Elijah Newrenf21ac362020-10-18 00:23:45 +0000318 $ sh ./t9200-git-cvsexport-commit.sh --run='!3,1-4'
Ilya Bobyr0445e6f2014-04-30 02:50:44 -0700319
320would just run tests from 1 to 4, including 3.
321
322You may use negation with ranges. The following will run all
323test in the test suite except from 7 up to 11:
324
325 $ sh ./t9200-git-cvsexport-commit.sh --run='!7-11'
326
Elijah Newrenf21ac362020-10-18 00:23:45 +0000327Sometimes there may be multiple tests with e.g. "setup" in their name
328that are needed and rather than figuring out the number for all of them
329we can just use "setup" as a substring/glob to match against the test
330description:
331
332 $ sh ./t0050-filesystem.sh --run=setup,9-11
333
334or one could select both the setup tests and the rename ones (assuming all
335relevant tests had those words in their descriptions):
336
337 $ sh ./t0050-filesystem.sh --run=setup,rename
338
Ilya Bobyr0445e6f2014-04-30 02:50:44 -0700339Some tests in a test suite rely on the previous tests performing
340certain actions, specifically some tests are designated as
341"setup" test, so you cannot _arbitrarily_ disable one test and
342expect the rest to function correctly.
343
344--run is mostly useful when you want to focus on a specific test
345and know what setup is needed for it. Or when you want to run
346everything up to a certain test.
Jakub Narebskifbd458a2008-06-20 23:10:50 +0200347
348
Nguyễn Thái Ngọc Duy4c2db932018-04-14 17:34:59 +0200349Running tests with special setups
350---------------------------------
351
352The whole test suite could be run to test some special features
353that cannot be easily covered by a few specific test cases. These
354could be enabled by running the test suite with correct GIT_TEST_
355environment set.
356
Ævar Arnfjörð Bjarmasonc7400392019-06-21 12:18:12 +0200357GIT_TEST_FAIL_PREREQS=<boolean> fails all prerequisites. This is
Ævar Arnfjörð Bjarmasondfe1a172019-05-13 20:32:42 +0200358useful for discovering issues with the tests where say a later test
359implicitly depends on an optional earlier test.
360
361There's a "FAIL_PREREQS" prerequisite that can be used to test for
362whether this mode is active, and e.g. skip some tests that are hard to
363refactor to deal with it. The "SYMLINKS" prerequisite is currently
364excluded as so much relies on it, but this might change in the future.
365
Nguyễn Thái Ngọc Duy4c2db932018-04-14 17:34:59 +0200366GIT_TEST_SPLIT_INDEX=<boolean> forces split-index mode on the whole
367test suite. Accept any boolean values that are accepted by git-config.
368
Jonathan Nieder33166f32019-12-23 17:02:28 -0800369GIT_TEST_PROTOCOL_VERSION=<n>, when set, makes 'protocol.version'
370default to n.
Jonathan Tan8cbeba02019-02-25 13:54:06 -0800371
Nguyễn Thái Ngọc Duy43fa44f2018-04-14 17:35:05 +0200372GIT_TEST_FULL_IN_PACK_ARRAY=<boolean> exercises the uncommon
373pack-objects code path where there are more than 1024 packs even if
374the actual number of packs in repository is below this limit. Accept
375any boolean values that are accepted by git-config.
376
Nguyễn Thái Ngọc Duyac77d0c2018-04-14 17:35:10 +0200377GIT_TEST_OE_SIZE=<n> exercises the uncommon pack-objects code path
378where we do not cache object size in memory and read it from existing
379packs on demand. This normally only happens when the object size is
380over 2GB. This variable forces the code path on any object larger than
381<n> bytes.
382
Ben Peartac6e12f2018-09-18 23:29:34 +0000383GIT_TEST_OE_DELTA_SIZE=<n> exercises the uncommon pack-objects code
Nguyễn Thái Ngọc Duy9ac3f0e2018-07-22 10:04:21 +0200384path where deltas larger than this limit require extra memory
385allocation for bookkeeping.
386
Nguyễn Thái Ngọc Duy5f4436a2018-08-25 15:02:09 +0200387GIT_TEST_VALIDATE_INDEX_CACHE_ENTRIES=<boolean> checks that cache-tree
388records are valid when the index is written out or after a merge. This
389is mostly to catch missing invalidation. Default is true.
390
Derrick Stolee859fdc02018-08-29 05:49:04 -0700391GIT_TEST_COMMIT_GRAPH=<boolean>, when true, forces the commit-graph to
392be written after every 'git commit' command, and overrides the
393'core.commitGraph' setting to true.
394
Garima Singhd5b873c2020-04-06 16:59:55 +0000395GIT_TEST_COMMIT_GRAPH_CHANGED_PATHS=<boolean>, when true, forces
396commit-graph write to compute and write changed path Bloom filters for
397every 'git commit-graph write', as if the `--changed-paths` option was
398passed in.
399
Ben Peart4cb54d02018-09-18 23:29:35 +0000400GIT_TEST_FSMONITOR=$PWD/t7519/fsmonitor-all exercises the fsmonitor
401code path for utilizing a file system monitor to speed up detecting
402new or changed files.
403
Ben Peart1f357b02018-09-18 23:29:36 +0000404GIT_TEST_INDEX_VERSION=<n> exercises the index read/write code path
405for the index version specified. Can be set to any valid version
406(currently 2, 3, or 4).
407
Derrick Stolee2d657ab2020-03-20 12:38:10 +0000408GIT_TEST_PACK_SPARSE=<boolean> if disabled will default the pack-objects
409builtin to use the non-sparse object walk. This can still be overridden by
410the --sparse command-line argument.
Derrick Stolee99dbbfa2019-01-16 10:26:01 -0800411
Ben Peart5765d972018-09-18 23:29:37 +0000412GIT_TEST_PRELOAD_INDEX=<boolean> exercises the preload-index code path
413by overriding the minimum number of cache entries required per thread.
414
Johannes Schindelinf83dff62019-11-13 12:40:57 +0000415GIT_TEST_ADD_I_USE_BUILTIN=<boolean>, when true, enables the
416built-in version of git add -i. See 'add.interactive.useBuiltin' in
417git-config(1).
418
Ben Peartc780b9c2018-10-10 11:59:35 -0400419GIT_TEST_INDEX_THREADS=<n> enables exercising the multi-threaded loading
420of the index for the whole test suite by bypassing the default number of
421cache entries and thread minimums. Setting this to 1 will make the
422index loading single threaded.
423
Derrick Stolee0465a502018-10-12 10:34:20 -0700424GIT_TEST_MULTI_PACK_INDEX=<boolean>, when true, forces the multi-pack-
425index to be written after every 'git repack' command, and overrides the
426'core.multiPackIndex' setting to true.
427
Jonathan Tan07c3c2a2019-01-16 11:28:15 -0800428GIT_TEST_SIDEBAND_ALL=<boolean>, when true, overrides the
429'uploadpack.allowSidebandAll' setting to true, and when false, forces
430fetch-pack to not request sideband-all (even if the server advertises
431sideband-all).
432
Johannes Schindelinb02e7d52019-04-12 02:37:24 -0700433GIT_TEST_DISALLOW_ABBREVIATED_OPTIONS=<boolean>, when true (which is
434the default when running tests), errors out when an abbreviated option
435is used.
436
Derrick Stoleef3d66ec2020-08-17 14:04:46 +0000437GIT_TEST_DEFAULT_HASH=<hash-algo> specifies which hash algorithm to
438use in the test scripts. Recognized values for <hash-algo> are "sha1"
439and "sha256".
440
Taylor Blaue8c58f82021-01-25 18:37:42 -0500441GIT_TEST_WRITE_REV_INDEX=<boolean>, when true enables the
442'pack.writeReverseIndex' setting.
443
Derrick Stoleeecfc47c2021-03-30 13:10:49 +0000444GIT_TEST_SPARSE_INDEX=<boolean>, when true enables index writes to use the
445sparse-index format by default.
446
Matheus Tavares87094fc2021-05-04 13:27:35 -0300447GIT_TEST_CHECKOUT_WORKERS=<n> overrides the 'checkout.workers' setting
448to <n> and 'checkout.thresholdForParallelism' to 0, forcing the
449execution of the parallel-checkout code.
450
Petr Baudisf50c9f72005-05-15 01:34:22 +0200451Naming Tests
452------------
453
454The test files are named as:
455
456 tNNNN-commandname-details.sh
457
458where N is a decimal digit.
459
460First digit tells the family:
461
462 0 - the absolute basics and global stuff
463 1 - the basic commands concerning database
464 2 - the basic commands concerning the working tree
465 3 - the other basic commands (e.g. ls-files)
466 4 - the diff commands
467 5 - the pull and exporting commands
468 6 - the revision tree commands (even e.g. merge-base)
Junio C Hamano8f4a9b62006-06-28 11:45:52 -0700469 7 - the porcelainish commands concerning the working tree
Jakub Narebski87577492006-12-29 14:39:09 +0100470 8 - the porcelainish commands concerning forensics
471 9 - the git tools
Petr Baudisf50c9f72005-05-15 01:34:22 +0200472
473Second digit tells the particular command we are testing.
474
475Third digit (optionally) tells the particular switch or group of switches
476we are testing.
477
Junio C Hamano77656602005-07-07 11:39:10 -0700478If you create files under t/ directory (i.e. here) that is not
479the top-level test script, never name the file to match the above
480pattern. The Makefile here considers all such files as the
Michael Witten63d32942011-02-22 17:15:00 +0000481top-level test script and tries to run all of them. Care is
Junio C Hamano77656602005-07-07 11:39:10 -0700482especially needed if you are creating a common test library
483file, similar to test-lib.sh, because such a library file may
484not be suitable for standalone execution.
485
Petr Baudisf50c9f72005-05-15 01:34:22 +0200486
Junio C Hamano986aa7f2005-05-14 00:25:06 -0700487Writing Tests
488-------------
489
490The test script is written as a shell script. It should start
Thomas Gummerer51b7a522017-11-26 20:20:59 +0000491with the standard "#!/bin/sh", and an
Junio C Hamano986aa7f2005-05-14 00:25:06 -0700492assignment to variable 'test_description', like this:
493
494 #!/bin/sh
Junio C Hamano986aa7f2005-05-14 00:25:06 -0700495
Junio C Hamano14cd1ff2005-05-15 14:21:13 -0700496 test_description='xxx test (option --frotz)
Junio C Hamano986aa7f2005-05-14 00:25:06 -0700497
498 This test registers the following structure in the cache
499 and tries to run git-ls-files with option --frotz.'
500
Petr Baudisf50c9f72005-05-15 01:34:22 +0200501
Junio C Hamano986aa7f2005-05-14 00:25:06 -0700502Source 'test-lib.sh'
503--------------------
504
505After assigning test_description, the test script should source
506test-lib.sh like this:
507
508 . ./test-lib.sh
509
510This test harness library does the following things:
511
512 - If the script is invoked with command line argument --help
513 (or -h), it shows the test_description and exits.
514
Ævar Arnfjörð Bjarmasone1ca1c92010-07-02 14:59:43 +0000515 - Creates an empty test directory with an empty .git/objects database
516 and chdir(2) into it. This directory is 't/trash
517 directory.$test_name_without_dotsh', with t/ subject to change by
SZEDER Gáborfb7d1e32019-01-05 02:08:59 +0100518 the --root option documented above, and a '.stress-<N>' suffix
519 appended by the --stress option.
Junio C Hamano986aa7f2005-05-14 00:25:06 -0700520
521 - Defines standard test helper functions for your scripts to
522 use. These functions are designed to make all scripts behave
523 consistently when command line arguments --verbose (or -v),
524 --debug (or -d), and --immediate (or -i) is given.
525
Matthew DeVore441ee352018-10-05 14:54:01 -0700526Do's & don'ts
527-------------
Ævar Arnfjörð Bjarmason20873f42010-07-02 14:59:49 +0000528
Junio C Hamano6fd45292010-07-05 11:37:30 -0700529Here are a few examples of things you probably should and shouldn't do
Ævar Arnfjörð Bjarmason20873f42010-07-02 14:59:49 +0000530when writing tests.
531
Matthew DeVore441ee352018-10-05 14:54:01 -0700532Here are the "do's:"
Ævar Arnfjörð Bjarmason20873f42010-07-02 14:59:49 +0000533
Junio C Hamano6fd45292010-07-05 11:37:30 -0700534 - Put all code inside test_expect_success and other assertions.
Ævar Arnfjörð Bjarmason20873f42010-07-02 14:59:49 +0000535
536 Even code that isn't a test per se, but merely some setup code
Junio C Hamano6fd45292010-07-05 11:37:30 -0700537 should be inside a test assertion.
Ævar Arnfjörð Bjarmason20873f42010-07-02 14:59:49 +0000538
539 - Chain your test assertions
540
541 Write test code like this:
542
543 git merge foo &&
544 git push bar &&
545 test ...
546
547 Instead of:
548
549 git merge hla
550 git push gh
551 test ...
552
553 That way all of the commands in your tests will succeed or fail. If
Elijah Newren00648ba2010-10-03 14:00:14 -0600554 you must ignore the return value of something, consider using a
555 helper function (e.g. use sane_unset instead of unset, in order
556 to avoid unportable return value for unsetting a variable that was
557 already unset), or prepending the command with test_might_fail or
558 test_must_fail.
Ævar Arnfjörð Bjarmason20873f42010-07-02 14:59:49 +0000559
Ævar Arnfjörð Bjarmason0c357542010-07-25 19:52:44 +0000560 - Check the test coverage for your tests. See the "Test coverage"
561 below.
562
Michael Witten63d32942011-02-22 17:15:00 +0000563 Don't blindly follow test coverage metrics; if a new function you added
564 doesn't have any coverage, then you're probably doing something wrong,
Ævar Arnfjörð Bjarmasone8b55f52010-07-25 19:52:45 +0000565 but having 100% coverage doesn't necessarily mean that you tested
566 everything.
567
568 Tests that are likely to smoke out future regressions are better
569 than tests that just inflate the coverage metrics.
570
Johannes Sixt95b104c2011-01-11 08:44:30 +0100571 - When a test checks for an absolute path that a git command generated,
572 construct the expected value using $(pwd) rather than $PWD,
573 $TEST_DIRECTORY, or $TRASH_DIRECTORY. It makes a difference on
574 Windows, where the shell (MSYS bash) mangles absolute path names.
575 For details, see the commit message of 4114156ae9.
576
Matthew DeVore441ee352018-10-05 14:54:01 -0700577 - Remember that inside the <script> part, the standard output and
578 standard error streams are discarded, and the test harness only
579 reports "ok" or "not ok" to the end user running the tests. Under
580 --verbose, they are shown to help debug the tests.
Ævar Arnfjörð Bjarmason20873f42010-07-02 14:59:49 +0000581
Junio C Hamano7cc112d2020-03-27 10:55:09 -0700582 - Be careful when you loop
583
584 You may need to verify multiple things in a loop, but the
585 following does not work correctly:
586
587 test_expect_success 'test three things' '
588 for i in one two three
589 do
590 test_something "$i"
591 done &&
592 test_something_else
593 '
594
595 Because the status of the loop itself is the exit status of the
596 test_something in the last round, the loop does not fail when
597 "test_something" for "one" or "two" fails. This is not what you
598 want.
599
600 Instead, you can break out of the loop immediately when you see a
601 failure. Because all test_expect_* snippets are executed inside
602 a function, "return 1" can be used to fail the test immediately
603 upon a failure:
604
605 test_expect_success 'test three things' '
606 for i in one two three
607 do
608 test_something "$i" || return 1
609 done &&
610 test_something_else
611 '
612
613 Note that we still &&-chain the loop to propagate failures from
614 earlier commands.
615
616
Matthew DeVore441ee352018-10-05 14:54:01 -0700617And here are the "don'ts:"
618
619 - Don't exit() within a <script> part.
Ævar Arnfjörð Bjarmason20873f42010-07-02 14:59:49 +0000620
621 The harness will catch this as a programming error of the test.
622 Use test_done instead if you need to stop the tests early (see
623 "Skipping tests" below).
624
Matthew DeVore441ee352018-10-05 14:54:01 -0700625 - Don't use '! git cmd' when you want to make sure the git command
626 exits with failure in a controlled way by calling "die()". Instead,
Junio C Hamanoad785852012-06-12 09:44:56 -0700627 use 'test_must_fail git cmd'. This will signal a failure if git
628 dies in an unexpected way (e.g. segfault).
629
Junio C Hamanof4455002013-06-04 09:50:12 -0700630 On the other hand, don't use test_must_fail for running regular
Junio C Hamano53de7422014-11-24 09:47:07 -0800631 platform commands; just use '! cmd'. We are not in the business
632 of verifying that the world given to us sanely works.
Junio C Hamanof4455002013-06-04 09:50:12 -0700633
Matthew DeVorea378fee2018-10-05 14:54:02 -0700634 - Don't feed the output of a git command to a pipe, as in:
635
636 git -C repo ls-files |
637 xargs -n 1 basename |
638 grep foo
639
640 which will discard git's exit code and may mask a crash. In the
641 above example, all exit codes are ignored except grep's.
642
643 Instead, write the output of that command to a temporary
644 file with ">" or assign it to a variable with "x=$(git ...)" rather
645 than pipe it.
646
647 - Don't use command substitution in a way that discards git's exit
648 code. When assigning to a variable, the exit code is not discarded,
649 e.g.:
650
651 x=$(git cat-file -p $sha) &&
652 ...
653
654 is OK because a crash in "git cat-file" will cause the "&&" chain
655 to fail, but:
656
657 test "refs/heads/foo" = "$(git symbolic-ref HEAD)"
658
659 is not OK and a crash in git could go undetected.
660
Matthew DeVore441ee352018-10-05 14:54:01 -0700661 - Don't use perl without spelling it as "$PERL_PATH". This is to help
662 our friends on Windows where the platform Perl often adds CR before
Junio C Hamanoad785852012-06-12 09:44:56 -0700663 the end of line, and they bundle Git with a version of Perl that
Jeff Kinga0e0ec92013-10-28 21:22:07 -0400664 does not do so, whose path is specified with $PERL_PATH. Note that we
665 provide a "perl" function which uses $PERL_PATH under the hood, so
666 you do not need to worry when simply running perl in the test scripts
667 (but you do, for example, on a shebang line or in a sub script
668 created via "write_script").
Junio C Hamanoad785852012-06-12 09:44:56 -0700669
Matthew DeVore441ee352018-10-05 14:54:01 -0700670 - Don't use sh without spelling it as "$SHELL_PATH", when the script
671 can be misinterpreted by broken platform shell (e.g. Solaris).
Junio C Hamanoad785852012-06-12 09:44:56 -0700672
Matthew DeVore441ee352018-10-05 14:54:01 -0700673 - Don't chdir around in tests. It is not sufficient to chdir to
Junio C Hamanoad785852012-06-12 09:44:56 -0700674 somewhere and then chdir back to the original location later in
675 the test, as any intermediate step can fail and abort the test,
676 causing the next test to start in an unexpected directory. Do so
677 inside a subshell if necessary.
678
Matthew DeVore441ee352018-10-05 14:54:01 -0700679 - Don't save and verify the standard error of compound commands, i.e.
680 group commands, subshells, and shell functions (except test helper
SZEDER Gábor94201a22018-02-24 00:39:50 +0100681 functions like 'test_must_fail') like this:
682
683 ( cd dir && git cmd ) 2>error &&
684 test_cmp expect error
685
686 When running the test with '-x' tracing, then the trace of commands
687 executed in the compound command will be included in standard error
688 as well, quite possibly throwing off the subsequent checks examining
689 the output. Instead, save only the relevant git command's standard
690 error:
691
692 ( cd dir && git cmd 2>../error ) &&
693 test_cmp expect error
694
Matthew DeVore441ee352018-10-05 14:54:01 -0700695 - Don't break the TAP output
Ævar Arnfjörð Bjarmason20873f42010-07-02 14:59:49 +0000696
Junio C Hamano6fd45292010-07-05 11:37:30 -0700697 The raw output from your test may be interpreted by a TAP harness. TAP
698 harnesses will ignore everything they don't know about, but don't step
699 on their toes in these areas:
Ævar Arnfjörð Bjarmason20873f42010-07-02 14:59:49 +0000700
701 - Don't print lines like "$x..$y" where $x and $y are integers.
702
703 - Don't print lines that begin with "ok" or "not ok".
704
Junio C Hamano6fd45292010-07-05 11:37:30 -0700705 TAP harnesses expect a line that begins with either "ok" and "not
Ævar Arnfjörð Bjarmason20873f42010-07-02 14:59:49 +0000706 ok" to signal a test passed or failed (and our harness already
707 produces such lines), so your script shouldn't emit such lines to
708 their output.
709
710 You can glean some further possible issues from the TAP grammar
Ævar Arnfjörð Bjarmasonc1d44ce2017-03-22 22:18:52 +0000711 (see https://metacpan.org/pod/TAP::Parser::Grammar#TAP-GRAMMAR)
Ævar Arnfjörð Bjarmason20873f42010-07-02 14:59:49 +0000712 but the best indication is to just run the tests with prove(1),
713 it'll complain if anything is amiss.
714
Ævar Arnfjörð Bjarmason20873f42010-07-02 14:59:49 +0000715
Ævar Arnfjörð Bjarmasonb5500d12010-07-02 14:59:48 +0000716Skipping tests
717--------------
718
Mathias Lafeldt681186a2011-03-09 21:25:09 +0100719If you need to skip tests you should do so by using the three-arg form
Ævar Arnfjörð Bjarmason99d90502010-07-28 10:34:59 +0000720of the test_* functions (see the "Test harness library" section
721below), e.g.:
722
Junio C Hamanoad785852012-06-12 09:44:56 -0700723 test_expect_success PERL 'I need Perl' '
Jeff Kinga0e0ec92013-10-28 21:22:07 -0400724 perl -e "hlagh() if unf_unf()"
Junio C Hamanoad785852012-06-12 09:44:56 -0700725 '
Ævar Arnfjörð Bjarmason99d90502010-07-28 10:34:59 +0000726
727The advantage of skipping tests like this is that platforms that don't
728have the PERL and other optional dependencies get an indication of how
729many tests they're missing.
730
731If the test code is too hairy for that (i.e. does a lot of setup work
732outside test assertions) you can also skip all remaining tests by
733setting skip_all and immediately call test_done:
Ævar Arnfjörð Bjarmasonb5500d12010-07-02 14:59:48 +0000734
735 if ! test_have_prereq PERL
736 then
737 skip_all='skipping perl interface tests, perl not available'
738 test_done
739 fi
Junio C Hamano14cd1ff2005-05-15 14:21:13 -0700740
Ævar Arnfjörð Bjarmason99d90502010-07-28 10:34:59 +0000741The string you give to skip_all will be used as an explanation for why
742the test was skipped.
743
Junio C Hamano986aa7f2005-05-14 00:25:06 -0700744End with test_done
745------------------
746
747Your script will be a sequence of tests, using helper functions
748from the test harness library. At the end of the script, call
749'test_done'.
750
751
752Test harness library
753--------------------
754
755There are a handful helper functions defined in the test harness
756library for your script to use.
757
Ævar Arnfjörð Bjarmason9a897892010-07-02 14:59:45 +0000758 - test_expect_success [<prereq>] <message> <script>
Junio C Hamano986aa7f2005-05-14 00:25:06 -0700759
Mathias Lafeldt72942a62011-04-26 12:33:26 +0200760 Usually takes two strings as parameters, and evaluates the
Junio C Hamano986aa7f2005-05-14 00:25:06 -0700761 <script>. If it yields success, test is considered
762 successful. <message> should state what it is testing.
763
764 Example:
765
766 test_expect_success \
767 'git-write-tree should be able to write an empty tree.' \
768 'tree=$(git-write-tree)'
769
Ævar Arnfjörð Bjarmason9a897892010-07-02 14:59:45 +0000770 If you supply three parameters the first will be taken to be a
Mathias Lafeldt72942a62011-04-26 12:33:26 +0200771 prerequisite; see the test_set_prereq and test_have_prereq
Ævar Arnfjörð Bjarmason9a897892010-07-02 14:59:45 +0000772 documentation below:
773
774 test_expect_success TTY 'git --paginate rev-list uses a pager' \
775 ' ... '
776
Ævar Arnfjörð Bjarmason93a57242010-08-06 21:19:23 +0000777 You can also supply a comma-separated list of prerequisites, in the
778 rare case where your test depends on more than one:
779
780 test_expect_success PERL,PYTHON 'yo dawg' \
781 ' test $(perl -E 'print eval "1 +" . qx[python -c "print 2"]') == "4" '
782
Ævar Arnfjörð Bjarmason9a897892010-07-02 14:59:45 +0000783 - test_expect_failure [<prereq>] <message> <script>
Junio C Hamano986aa7f2005-05-14 00:25:06 -0700784
Junio C Hamano41ac4142008-02-01 01:50:53 -0800785 This is NOT the opposite of test_expect_success, but is used
786 to mark a test that demonstrates a known breakage. Unlike
787 the usual test_expect_success tests, which say "ok" on
788 success and "FAIL" on failure, this will say "FIXED" on
789 success and "still broken" on failure. Failures from these
790 tests won't cause -i (immediate) to stop.
Junio C Hamano986aa7f2005-05-14 00:25:06 -0700791
Ævar Arnfjörð Bjarmason9a897892010-07-02 14:59:45 +0000792 Like test_expect_success this function can optionally use a three
793 argument invocation with a prerequisite as the first argument.
794
Junio C Hamano986aa7f2005-05-14 00:25:06 -0700795 - test_debug <script>
796
797 This takes a single argument, <script>, and evaluates it only
798 when the test script is started with --debug command line
799 argument. This is primarily meant for use during the
800 development of a new test script.
801
Johannes Schindelin6a940882015-10-30 12:02:56 -0700802 - debug <git-command>
803
804 Run a git command inside a debugger. This is primarily meant for
805 use when debugging a failing test script.
806
Junio C Hamano986aa7f2005-05-14 00:25:06 -0700807 - test_done
808
809 Your test script must have test_done at the end. Its purpose
810 is to summarize successes and failures in the test script and
811 exit with an appropriate error code.
812
Johannes Schindelin00884962009-01-27 23:34:48 +0100813 - test_tick
814
815 Make commit and tag names consistent by setting the author and
Michael Witten63d32942011-02-22 17:15:00 +0000816 committer times to defined state. Subsequent calls will
Johannes Schindelin00884962009-01-27 23:34:48 +0100817 advance the times by a fixed amount.
818
819 - test_commit <message> [<filename> [<contents>]]
820
821 Creates a commit with the given message, committing the given
822 file with the given contents (default for both is to reuse the
823 message string), and adds a tag (again reusing the message
824 string as name). Calls test_tick to make the SHA-1s
825 reproducible.
826
827 - test_merge <message> <commit-or-tag>
828
829 Merges the given rev using the given message. Like test_commit,
830 creates a tag and calls test_tick before committing.
Junio C Hamano986aa7f2005-05-14 00:25:06 -0700831
Mathias Lafeldt72942a62011-04-26 12:33:26 +0200832 - test_set_prereq <prereq>
Ævar Arnfjörð Bjarmason9a897892010-07-02 14:59:45 +0000833
834 Set a test prerequisite to be used later with test_have_prereq. The
Ævar Arnfjörð Bjarmasonbe53dee2010-08-06 21:19:25 +0000835 test-lib will set some prerequisites for you, see the
836 "Prerequisites" section below for a full list of these.
837
838 Others you can set yourself and use later with either
839 test_have_prereq directly, or the three argument invocation of
840 test_expect_success and test_expect_failure.
Ævar Arnfjörð Bjarmason9a897892010-07-02 14:59:45 +0000841
Mathias Lafeldt72942a62011-04-26 12:33:26 +0200842 - test_have_prereq <prereq>
Ævar Arnfjörð Bjarmason9a897892010-07-02 14:59:45 +0000843
Ævar Arnfjörð Bjarmason44730602017-03-22 22:18:54 +0000844 Check if we have a prerequisite previously set with test_set_prereq.
845 The most common way to use this explicitly (as opposed to the
846 implicit use when an argument is passed to test_expect_*) is to skip
847 all the tests at the start of the test script if we don't have some
848 essential prerequisite:
Ævar Arnfjörð Bjarmason9a897892010-07-02 14:59:45 +0000849
850 if ! test_have_prereq PERL
851 then
852 skip_all='skipping perl interface tests, perl not available'
853 test_done
854 fi
855
Ævar Arnfjörð Bjarmason2fac6a42010-07-02 14:59:46 +0000856 - test_external [<prereq>] <message> <external> <script>
857
858 Execute a <script> with an <external> interpreter (like perl). This
859 was added for tests like t9700-perl-git.sh which do most of their
860 work in an external test script.
861
862 test_external \
863 'GitwebCache::*FileCache*' \
Jeff Kinga0e0ec92013-10-28 21:22:07 -0400864 perl "$TEST_DIRECTORY"/t9503/test_cache_interface.pl
Ævar Arnfjörð Bjarmason2fac6a42010-07-02 14:59:46 +0000865
866 If the test is outputting its own TAP you should set the
867 test_external_has_tap variable somewhere before calling the first
868 test_external* function. See t9700-perl-git.sh for an example.
869
870 # The external test will outputs its own plan
871 test_external_has_tap=1
872
873 - test_external_without_stderr [<prereq>] <message> <external> <script>
874
875 Like test_external but fail if there's any output on stderr,
876 instead of checking the exit code.
877
878 test_external_without_stderr \
879 'Perl API' \
Jeff Kinga0e0ec92013-10-28 21:22:07 -0400880 perl "$TEST_DIRECTORY"/t9700/test.pl
Ævar Arnfjörð Bjarmason2fac6a42010-07-02 14:59:46 +0000881
Ævar Arnfjörð Bjarmason892e6f72010-10-03 13:59:59 -0600882 - test_expect_code <exit-code> <command>
883
884 Run a command and ensure that it exits with the given exit code.
885 For example:
886
887 test_expect_success 'Merge with d/f conflicts' '
888 test_expect_code 1 git merge "merge msg" B master
889 '
890
SZEDER Gábor12e31a62018-02-09 03:42:33 +0100891 - test_must_fail [<options>] <git-command>
Jonathan Niederc9667452010-07-06 15:04:10 -0500892
893 Run a git command and ensure it fails in a controlled way. Use
Brandon Casey971ecbd2010-07-20 12:17:12 -0500894 this instead of "! <git-command>". When git-command dies due to a
895 segfault, test_must_fail diagnoses it as an error; "! <git-command>"
896 treats it as just another expected failure, which would let such a
897 bug go unnoticed.
Jonathan Niederc9667452010-07-06 15:04:10 -0500898
SZEDER Gábor12e31a62018-02-09 03:42:33 +0100899 Accepts the following options:
900
901 ok=<signal-name>[,<...>]:
902 Don't treat an exit caused by the given signal as error.
903 Multiple signals can be specified as a comma separated list.
904 Currently recognized signal names are: sigpipe, success.
905 (Don't use 'success', use 'test_might_fail' instead.)
906
907 - test_might_fail [<options>] <git-command>
Jonathan Niederc9667452010-07-06 15:04:10 -0500908
909 Similar to test_must_fail, but tolerate success, too. Use this
910 instead of "<git-command> || :" to catch failures due to segv.
911
SZEDER Gábor12e31a62018-02-09 03:42:33 +0100912 Accepts the same options as test_must_fail.
913
Jonathan Niederc9667452010-07-06 15:04:10 -0500914 - test_cmp <expected> <actual>
915
916 Check whether the content of the <actual> file matches the
917 <expected> file. This behaves like "cmp" but produces more
918 helpful output when the test is run with "-v" option.
919
Thomas Gummerer5a052622017-11-26 20:21:00 +0000920 - test_cmp_rev <expected> <actual>
921
922 Check whether the <expected> rev points to the same commit as the
923 <actual> rev.
924
Jonathan Niederfb3340a2010-10-31 02:33:50 -0500925 - test_line_count (= | -lt | -ge | ...) <length> <file>
926
927 Check whether a file has the length it is expected to.
928
Ævar Arnfjörð Bjarmason45a26862021-02-12 14:29:41 +0100929 - test_path_is_file <path>
930 test_path_is_dir <path>
931 test_path_is_missing <path>
Matthieu Moy2caf20c2010-08-10 17:17:52 +0200932
Mathias Lafeldt72942a62011-04-26 12:33:26 +0200933 Check if the named path is a file, if the named path is a
934 directory, or if the named path does not exist, respectively,
Ævar Arnfjörð Bjarmason45a26862021-02-12 14:29:41 +0100935 and fail otherwise.
Matthieu Moy2caf20c2010-08-10 17:17:52 +0200936
Jonathan Niederc9667452010-07-06 15:04:10 -0500937 - test_when_finished <script>
938
939 Prepend <script> to a list of commands to run to clean up
940 at the end of the current test. If some clean-up command
941 fails, the test will not pass.
942
943 Example:
944
945 test_expect_success 'branch pointing to non-commit' '
946 git rev-parse HEAD^{tree} >.git/refs/heads/invalid &&
947 test_when_finished "git update-ref -d refs/heads/invalid" &&
948 ...
949 '
950
Johannes Schindelin900721e2019-03-13 13:24:11 +0100951 - test_atexit <script>
952
953 Prepend <script> to a list of commands to run unconditionally to
954 clean up before the test script exits, e.g. to stop a daemon:
955
956 test_expect_success 'test git daemon' '
957 git daemon &
958 daemon_pid=$! &&
959 test_atexit 'kill $daemon_pid' &&
960 hello world
961 '
962
963 The commands will be executed before the trash directory is removed,
964 i.e. the atexit commands will still be able to access any pidfiles or
965 socket files.
966
967 Note that these commands will be run even when a test script run
968 with '--immediate' fails. Be careful with your atexit commands to
969 minimize any changes to the failed state.
970
Jonathan Niederbb98b012014-05-05 16:51:43 -0700971 - test_write_lines <lines>
Michael S. Tsirkinac9afcc2014-04-27 21:15:47 +0300972
Jonathan Niederbb98b012014-05-05 16:51:43 -0700973 Write <lines> on standard output, one line per argument.
Michael S. Tsirkinac9afcc2014-04-27 21:15:47 +0300974 Useful to prepare multi-line files in a compact form.
975
976 Example:
977
Jonathan Niederbb98b012014-05-05 16:51:43 -0700978 test_write_lines a b c d e f g >foo
Michael S. Tsirkinac9afcc2014-04-27 21:15:47 +0300979
980 Is a more compact equivalent of:
981 cat >foo <<-EOF
982 a
983 b
984 c
985 d
986 e
987 f
988 g
989 EOF
990
991
Jens Lehmannc4d25392012-01-17 22:04:31 +0100992 - test_pause
993
994 This command is useful for writing and debugging tests and must be
995 removed before submitting. It halts the execution of the test and
996 spawns a shell in the trash directory. Exit the shell to continue
997 the test. Example:
998
999 test_expect_success 'test' '
1000 git do-something >actual &&
1001 test_pause &&
1002 test_cmp expected actual
1003 '
1004
Johannes Sixt9ce415d2013-06-07 22:53:27 +02001005 - test_ln_s_add <path1> <path2>
1006
1007 This function helps systems whose filesystem does not support symbolic
1008 links. Use it to add a symbolic link entry to the index when it is not
1009 important that the file system entry is a symbolic link, i.e., instead
1010 of the sequence
1011
1012 ln -s foo bar &&
1013 git add bar
1014
1015 Sometimes it is possible to split a test in a part that does not need
1016 the symbolic link in the file system and a part that does; then only
1017 the latter part need be protected by a SYMLINKS prerequisite (see below).
1018
brian m. carlson2c02b112018-09-13 05:17:31 +00001019 - test_oid_init
1020
1021 This function loads facts and useful object IDs related to the hash
1022 algorithm(s) in use from the files in t/oid-info.
1023
1024 - test_oid_cache
1025
1026 This function reads per-hash algorithm information from standard
1027 input (usually a heredoc) in the format described in
1028 t/oid-info/README. This is useful for test-specific values, such as
1029 object IDs, which must vary based on the hash algorithm.
1030
1031 Certain fixed values, such as hash sizes and common placeholder
1032 object IDs, can be loaded with test_oid_init (described above).
1033
1034 - test_oid <key>
1035
1036 This function looks up a value for the hash algorithm in use, based
1037 on the key given. The value must have been loaded using
1038 test_oid_init or test_oid_cache. Providing an unknown key is an
1039 error.
1040
Junio C Hamano11f470a2019-02-09 10:25:26 -08001041 - yes [<string>]
1042
1043 This is often seen in modern UNIX but some platforms lack it, so
1044 the test harness overrides the platform implementation with a
1045 more limited one. Use this only when feeding a handful lines of
1046 output to the downstream---unlike the real version, it generates
1047 only up to 99 lines.
1048
SZEDER Gábor43a2afe2019-11-22 14:14:36 +01001049 - test_bool_env <env-variable-name> <default-value>
1050
1051 Given the name of an environment variable with a bool value,
1052 normalize its value to a 0 (true) or 1 (false or empty string)
1053 return code. Return with code corresponding to the given default
1054 value if the variable is unset.
1055 Abort the test script if either the value of the variable or the
1056 default are not valid bool values.
1057
Junio C Hamano11f470a2019-02-09 10:25:26 -08001058
Ævar Arnfjörð Bjarmasonbe53dee2010-08-06 21:19:25 +00001059Prerequisites
1060-------------
1061
1062These are the prerequisites that the test library predefines with
1063test_have_prereq.
1064
1065See the prereq argument to the test_* functions in the "Test harness
1066library" section above and the "test_have_prereq" function for how to
1067use these, and "test_set_prereq" for how to define your own.
1068
Jonathan Niederf8fc0ee2013-10-28 12:22:16 -07001069 - PYTHON
Ævar Arnfjörð Bjarmasonbe53dee2010-08-06 21:19:25 +00001070
Jonathan Niederf8fc0ee2013-10-28 12:22:16 -07001071 Git wasn't compiled with NO_PYTHON=YesPlease. Wrap any tests that
1072 need Python with this.
1073
1074 - PERL
1075
1076 Git wasn't compiled with NO_PERL=YesPlease.
1077
1078 Even without the PERL prerequisite, tests can assume there is a
1079 usable perl interpreter at $PERL_PATH, though it need not be
1080 particularly modern.
Ævar Arnfjörð Bjarmasonbe53dee2010-08-06 21:19:25 +00001081
1082 - POSIXPERM
1083
1084 The filesystem supports POSIX style permission bits.
1085
1086 - BSLASHPSPEC
1087
1088 Backslashes in pathspec are not directory separators. This is not
1089 set on Windows. See 6fd1106a for details.
1090
1091 - EXECKEEPSPID
1092
1093 The process retains the same pid across exec(2). See fb9a2bea for
1094 details.
1095
Adam Spiers20073272013-04-11 03:07:04 +01001096 - PIPE
1097
1098 The filesystem we're on supports creation of FIFOs (named pipes)
1099 via mkfifo(1).
1100
Ævar Arnfjörð Bjarmasonbe53dee2010-08-06 21:19:25 +00001101 - SYMLINKS
1102
1103 The filesystem we're on supports symbolic links. E.g. a FAT
1104 filesystem doesn't support these. See 704a3143 for details.
Ævar Arnfjörð Bjarmason2fac6a42010-07-02 14:59:46 +00001105
Ævar Arnfjörð Bjarmasonc91cfd12010-08-06 22:09:09 +00001106 - SANITY
1107
1108 Test is not run by root user, and an attempt to write to an
1109 unwritable file is expected to fail correctly.
Junio C Hamano986aa7f2005-05-14 00:25:06 -07001110
Ævar Arnfjörð Bjarmason3eb585c2017-05-20 21:42:06 +00001111 - PCRE
Michał Kiedrowicz8f852ce2011-05-09 23:52:07 +02001112
Ævar Arnfjörð Bjarmason3eb585c2017-05-20 21:42:06 +00001113 Git was compiled with support for PCRE. Wrap any tests
Michał Kiedrowicz8f852ce2011-05-09 23:52:07 +02001114 that use git-grep --perl-regexp or git-grep -P in these.
1115
Michael J Gruberac39aa62012-07-26 15:39:53 +02001116 - CASE_INSENSITIVE_FS
1117
1118 Test is run on a case insensitive file system.
1119
Michael J Gruber5b0b5dd2012-07-26 15:39:56 +02001120 - UTF8_NFD_TO_NFC
1121
1122 Test is run on a filesystem which converts decomposed utf-8 (nfd)
1123 to precomposed utf-8 (nfc).
1124
Ævar Arnfjörð Bjarmason68c7d272017-05-25 19:45:31 +00001125 - PTHREADS
1126
1127 Git wasn't compiled with NO_PTHREADS=YesPlease.
1128
Junio C Hamano986aa7f2005-05-14 00:25:06 -07001129Tips for Writing Tests
1130----------------------
1131
1132As with any programming projects, existing programs are the best
1133source of the information. However, do _not_ emulate
1134t0000-basic.sh when writing your tests. The test is special in
Denton Liu788db142020-06-07 05:48:26 -04001135that it tries to validate the very core of Git. For example, it
Junio C Hamano986aa7f2005-05-14 00:25:06 -07001136knows that there will be 256 subdirectories under .git/objects/,
1137and it knows that the object ID of an empty tree is a certain
113840-byte string. This is deliberately done so in t0000-basic.sh
1139because the things the very basic core test tries to achieve is
Denton Liu788db142020-06-07 05:48:26 -04001140to serve as a basis for people who are changing the Git internals
Junio C Hamano986aa7f2005-05-14 00:25:06 -07001141drastically. For these people, after making certain changes,
1142not seeing failures from the basic test _is_ a failure. And
Denton Liu788db142020-06-07 05:48:26 -04001143such drastic changes to the core Git that even changes these
Junio C Hamano986aa7f2005-05-14 00:25:06 -07001144otherwise supposedly stable object IDs should be accompanied by
1145an update to t0000-basic.sh.
1146
1147However, other tests that simply rely on basic parts of the core
Denton Liu788db142020-06-07 05:48:26 -04001148Git working properly should not have that level of intimate
1149knowledge of the core Git internals. If all the test scripts
Junio C Hamano986aa7f2005-05-14 00:25:06 -07001150hardcoded the object IDs like t0000-basic.sh does, that defeats
1151the purpose of t0000-basic.sh, which is to isolate that level of
1152validation in one place. Your test also ends up needing
1153updating when such a change to the internal happens, so do _not_
1154do it and leave the low level of validation to t0000-basic.sh.
Ævar Arnfjörð Bjarmasond15e9eb2010-08-08 14:49:25 +00001155
Ævar Arnfjörð Bjarmason0c357542010-07-25 19:52:44 +00001156Test coverage
1157-------------
1158
1159You can use the coverage tests to find code paths that are not being
1160used or properly exercised yet.
1161
1162To do that, run the coverage target at the top-level (not in the t/
1163directory):
1164
1165 make coverage
1166
1167That'll compile Git with GCC's coverage arguments, and generate a test
1168report with gcov after the tests finish. Running the coverage tests
1169can take a while, since running the tests in parallel is incompatible
1170with GCC's coverage mode.
1171
1172After the tests have run you can generate a list of untested
1173functions:
1174
1175 make coverage-untested-functions
1176
1177You can also generate a detailed per-file HTML report using the
1178Devel::Cover module. To install it do:
1179
1180 # On Debian or Ubuntu:
1181 sudo aptitude install libdevel-cover-perl
1182
1183 # From the CPAN with cpanminus
1184 curl -L http://cpanmin.us | perl - --sudo --self-upgrade
1185 cpanm --sudo Devel::Cover
1186
1187Then, at the top-level:
1188
1189 make cover_db_html
1190
1191That'll generate a detailed cover report in the "cover_db_html"
1192directory, which you can then copy to a webserver, or inspect locally
1193in a browser.