1Core GIT Tests 2============== 3 4This directory holds many test scripts for core GIT tools. The 5first 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 *** 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 24 ... 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) 29 1..43 30 *** t0001-init.sh *** 31 ok 1 - plain 32 ok 2 - plain with GIT_WORK_TREE 33 ok 3 - plain bare 34 35Since the tests all output TAP (see http://testanything.org) they can 36be run with any TAP harness. Here's an example of parallel testing 37powered by a recent version of prove(1): 38 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... )=== 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 53You 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 59You 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 69 70You 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 72appropriately before running "make". 73 74--verbose:: 75 This makes the test more verbose. Specifically, the 76 command being run and their output if any are also 77 output. 78 79--debug:: 80 This may help the person who is developing a new test. 81 It causes the command defined with test_debug to run. 82 The "trash" directory (used to store all temporary data 83 during testing) is not deleted even if there are no 84 failed tests so that you can inspect its contents after 85 the test finished. 86 87--immediate:: 88 This causes the test to immediately exit upon the first 89 failed test. 90 91--long-tests:: 92 This causes additional long-running tests to be run (where 93 available), for more exhaustive testing. 94 95--valgrind:: 96 Execute all Git binaries with valgrind and exit with status 97 126 on errors (just like regular tests, this will only stop 98 the test script when running under -i). Valgrind errors 99 go to stderr, so you might want to pass the -v option, too. 100 101 Since it makes no sense to run the tests with --valgrind and 102 not see any output, this option implies --verbose. For 103 convenience, it also implies --tee. 104 105 Note that valgrind is run with the option --leak-check=no, 106 as the git process is short-lived and some errors are not 107 interesting. In order to run a single command under the same 108 conditions manually, you should set GIT_VALGRIND to point to 109 the 't/valgrind/' directory and use the commands under 110 't/valgrind/bin/'. 111 112--tee:: 113 In addition to printing the test output to the terminal, 114 write it to files named 't/test-results/$TEST_NAME.out'. 115 As the names depend on the tests' file names, it is safe to 116 run the tests with this option in parallel. 117 118--with-dashes:: 119 By default tests are run without dashed forms of 120 commands (like git-commit) in the PATH (it only uses 121 wrappers from ../bin-wrappers). Use this option to include 122 the build directory (..) in the PATH, which contains all 123 the dashed forms of commands. This option is currently 124 implied by other options like --valgrind and 125 GIT_TEST_INSTALLED. 126 127--root=<directory>:: 128 Create "trash" directories used to store all temporary data during 129 testing under <directory>, instead of the t/ directory. 130 Using this option with a RAM-based filesystem (such as tmpfs) 131 can massively speed up the test suite. 132 133You can also set the GIT_TEST_INSTALLED environment variable to 134the bindir of an existing git installation to test that installation. 135You still need to have built this git sandbox, from which various 136test-* support programs, templates, and perl libraries are used. 137If your installed git is incomplete, it will silently test parts of 138your built version instead. 139 140When using GIT_TEST_INSTALLED, you can also set GIT_TEST_EXEC_PATH to 141override the location of the dashed-form subcommands (what 142GIT_EXEC_PATH would be used for during normal operation). 143GIT_TEST_EXEC_PATH defaults to `$GIT_TEST_INSTALLED/git --exec-path`. 144 145 146Skipping Tests 147-------------- 148 149In some environments, certain tests have no way of succeeding 150due to platform limitation, such as lack of 'unzip' program, or 151filesystem that do not allow arbitrary sequence of non-NUL bytes 152as pathnames. 153 154You should be able to say something like 155 156 $ GIT_SKIP_TESTS=t9200.8 sh ./t9200-git-cvsexport-commit.sh 157 158and even: 159 160 $ GIT_SKIP_TESTS='t[0-4]??? t91?? t9200.8' make 161 162to omit such tests. The value of the environment variable is a 163SP separated list of patterns that tells which tests to skip, 164and either can match the "t[0-9]{4}" part to skip the whole 165test, or t[0-9]{4} followed by ".$number" to say which 166particular test to skip. 167 168Note that some tests in the existing test suite rely on previous 169test item, so you cannot arbitrarily disable one and expect the 170remainder of test to check what the test originally was intended 171to check. 172 173 174Naming Tests 175------------ 176 177The test files are named as: 178 179 tNNNN-commandname-details.sh 180 181where N is a decimal digit. 182 183First digit tells the family: 184 185 0 - the absolute basics and global stuff 186 1 - the basic commands concerning database 187 2 - the basic commands concerning the working tree 188 3 - the other basic commands (e.g. ls-files) 189 4 - the diff commands 190 5 - the pull and exporting commands 191 6 - the revision tree commands (even e.g. merge-base) 192 7 - the porcelainish commands concerning the working tree 193 8 - the porcelainish commands concerning forensics 194 9 - the git tools 195 196Second digit tells the particular command we are testing. 197 198Third digit (optionally) tells the particular switch or group of switches 199we are testing. 200 201If you create files under t/ directory (i.e. here) that is not 202the top-level test script, never name the file to match the above 203pattern. The Makefile here considers all such files as the 204top-level test script and tries to run all of them. Care is 205especially needed if you are creating a common test library 206file, similar to test-lib.sh, because such a library file may 207not be suitable for standalone execution. 208 209 210Writing Tests 211------------- 212 213The test script is written as a shell script. It should start 214with the standard "#!/bin/sh" with copyright notices, and an 215assignment to variable 'test_description', like this: 216 217 #!/bin/sh 218 # 219 # Copyright (c) 2005 Junio C Hamano 220 # 221 222 test_description='xxx test (option --frotz) 223 224 This test registers the following structure in the cache 225 and tries to run git-ls-files with option --frotz.' 226 227 228Source 'test-lib.sh' 229-------------------- 230 231After assigning test_description, the test script should source 232test-lib.sh like this: 233 234 . ./test-lib.sh 235 236This test harness library does the following things: 237 238 - If the script is invoked with command line argument --help 239 (or -h), it shows the test_description and exits. 240 241 - Creates an empty test directory with an empty .git/objects database 242 and chdir(2) into it. This directory is 't/trash 243 directory.$test_name_without_dotsh', with t/ subject to change by 244 the --root option documented above. 245 246 - Defines standard test helper functions for your scripts to 247 use. These functions are designed to make all scripts behave 248 consistently when command line arguments --verbose (or -v), 249 --debug (or -d), and --immediate (or -i) is given. 250 251Do's, don'ts & things to keep in mind 252------------------------------------- 253 254Here are a few examples of things you probably should and shouldn't do 255when writing tests. 256 257Do: 258 259 - Put all code inside test_expect_success and other assertions. 260 261 Even code that isn't a test per se, but merely some setup code 262 should be inside a test assertion. 263 264 - Chain your test assertions 265 266 Write test code like this: 267 268 git merge foo && 269 git push bar && 270 test ... 271 272 Instead of: 273 274 git merge hla 275 git push gh 276 test ... 277 278 That way all of the commands in your tests will succeed or fail. If 279 you must ignore the return value of something, consider using a 280 helper function (e.g. use sane_unset instead of unset, in order 281 to avoid unportable return value for unsetting a variable that was 282 already unset), or prepending the command with test_might_fail or 283 test_must_fail. 284 285 - Check the test coverage for your tests. See the "Test coverage" 286 below. 287 288 Don't blindly follow test coverage metrics; if a new function you added 289 doesn't have any coverage, then you're probably doing something wrong, 290 but having 100% coverage doesn't necessarily mean that you tested 291 everything. 292 293 Tests that are likely to smoke out future regressions are better 294 than tests that just inflate the coverage metrics. 295 296 - When a test checks for an absolute path that a git command generated, 297 construct the expected value using $(pwd) rather than $PWD, 298 $TEST_DIRECTORY, or $TRASH_DIRECTORY. It makes a difference on 299 Windows, where the shell (MSYS bash) mangles absolute path names. 300 For details, see the commit message of 4114156ae9. 301 302Don't: 303 304 - exit() within a <script> part. 305 306 The harness will catch this as a programming error of the test. 307 Use test_done instead if you need to stop the tests early (see 308 "Skipping tests" below). 309 310 - use '! git cmd' when you want to make sure the git command exits 311 with failure in a controlled way by calling "die()". Instead, 312 use 'test_must_fail git cmd'. This will signal a failure if git 313 dies in an unexpected way (e.g. segfault). 314 315 - use perl without spelling it as "$PERL_PATH". This is to help our 316 friends on Windows where the platform Perl often adds CR before 317 the end of line, and they bundle Git with a version of Perl that 318 does not do so, whose path is specified with $PERL_PATH. 319 320 - use sh without spelling it as "$SHELL_PATH", when the script can 321 be misinterpreted by broken platform shell (e.g. Solaris). 322 323 - chdir around in tests. It is not sufficient to chdir to 324 somewhere and then chdir back to the original location later in 325 the test, as any intermediate step can fail and abort the test, 326 causing the next test to start in an unexpected directory. Do so 327 inside a subshell if necessary. 328 329 - Break the TAP output 330 331 The raw output from your test may be interpreted by a TAP harness. TAP 332 harnesses will ignore everything they don't know about, but don't step 333 on their toes in these areas: 334 335 - Don't print lines like "$x..$y" where $x and $y are integers. 336 337 - Don't print lines that begin with "ok" or "not ok". 338 339 TAP harnesses expect a line that begins with either "ok" and "not 340 ok" to signal a test passed or failed (and our harness already 341 produces such lines), so your script shouldn't emit such lines to 342 their output. 343 344 You can glean some further possible issues from the TAP grammar 345 (see http://search.cpan.org/perldoc?TAP::Parser::Grammar#TAP_Grammar) 346 but the best indication is to just run the tests with prove(1), 347 it'll complain if anything is amiss. 348 349Keep in mind: 350 351 - Inside <script> part, the standard output and standard error 352 streams are discarded, and the test harness only reports "ok" or 353 "not ok" to the end user running the tests. Under --verbose, they 354 are shown to help debugging the tests. 355 356 357Skipping tests 358-------------- 359 360If you need to skip tests you should do so by using the three-arg form 361of the test_* functions (see the "Test harness library" section 362below), e.g.: 363 364 test_expect_success PERL 'I need Perl' ' 365 "$PERL_PATH" -e "hlagh() if unf_unf()" 366 ' 367 368The advantage of skipping tests like this is that platforms that don't 369have the PERL and other optional dependencies get an indication of how 370many tests they're missing. 371 372If the test code is too hairy for that (i.e. does a lot of setup work 373outside test assertions) you can also skip all remaining tests by 374setting skip_all and immediately call test_done: 375 376 if ! test_have_prereq PERL 377 then 378 skip_all='skipping perl interface tests, perl not available' 379 test_done 380 fi 381 382The string you give to skip_all will be used as an explanation for why 383the test was skipped. 384 385End with test_done 386------------------ 387 388Your script will be a sequence of tests, using helper functions 389from the test harness library. At the end of the script, call 390'test_done'. 391 392 393Test harness library 394-------------------- 395 396There are a handful helper functions defined in the test harness 397library for your script to use. 398 399 - test_expect_success [<prereq>] <message> <script> 400 401 Usually takes two strings as parameters, and evaluates the 402 <script>. If it yields success, test is considered 403 successful. <message> should state what it is testing. 404 405 Example: 406 407 test_expect_success \ 408 'git-write-tree should be able to write an empty tree.' \ 409 'tree=$(git-write-tree)' 410 411 If you supply three parameters the first will be taken to be a 412 prerequisite; see the test_set_prereq and test_have_prereq 413 documentation below: 414 415 test_expect_success TTY 'git --paginate rev-list uses a pager' \ 416 ' ... ' 417 418 You can also supply a comma-separated list of prerequisites, in the 419 rare case where your test depends on more than one: 420 421 test_expect_success PERL,PYTHON 'yo dawg' \ 422 ' test $(perl -E 'print eval "1 +" . qx[python -c "print 2"]') == "4" ' 423 424 - test_expect_failure [<prereq>] <message> <script> 425 426 This is NOT the opposite of test_expect_success, but is used 427 to mark a test that demonstrates a known breakage. Unlike 428 the usual test_expect_success tests, which say "ok" on 429 success and "FAIL" on failure, this will say "FIXED" on 430 success and "still broken" on failure. Failures from these 431 tests won't cause -i (immediate) to stop. 432 433 Like test_expect_success this function can optionally use a three 434 argument invocation with a prerequisite as the first argument. 435 436 - test_debug <script> 437 438 This takes a single argument, <script>, and evaluates it only 439 when the test script is started with --debug command line 440 argument. This is primarily meant for use during the 441 development of a new test script. 442 443 - test_done 444 445 Your test script must have test_done at the end. Its purpose 446 is to summarize successes and failures in the test script and 447 exit with an appropriate error code. 448 449 - test_tick 450 451 Make commit and tag names consistent by setting the author and 452 committer times to defined state. Subsequent calls will 453 advance the times by a fixed amount. 454 455 - test_commit <message> [<filename> [<contents>]] 456 457 Creates a commit with the given message, committing the given 458 file with the given contents (default for both is to reuse the 459 message string), and adds a tag (again reusing the message 460 string as name). Calls test_tick to make the SHA-1s 461 reproducible. 462 463 - test_merge <message> <commit-or-tag> 464 465 Merges the given rev using the given message. Like test_commit, 466 creates a tag and calls test_tick before committing. 467 468 - test_set_prereq <prereq> 469 470 Set a test prerequisite to be used later with test_have_prereq. The 471 test-lib will set some prerequisites for you, see the 472 "Prerequisites" section below for a full list of these. 473 474 Others you can set yourself and use later with either 475 test_have_prereq directly, or the three argument invocation of 476 test_expect_success and test_expect_failure. 477 478 - test_have_prereq <prereq> 479 480 Check if we have a prerequisite previously set with 481 test_set_prereq. The most common use of this directly is to skip 482 all the tests if we don't have some essential prerequisite: 483 484 if ! test_have_prereq PERL 485 then 486 skip_all='skipping perl interface tests, perl not available' 487 test_done 488 fi 489 490 - test_external [<prereq>] <message> <external> <script> 491 492 Execute a <script> with an <external> interpreter (like perl). This 493 was added for tests like t9700-perl-git.sh which do most of their 494 work in an external test script. 495 496 test_external \ 497 'GitwebCache::*FileCache*' \ 498 "$PERL_PATH" "$TEST_DIRECTORY"/t9503/test_cache_interface.pl 499 500 If the test is outputting its own TAP you should set the 501 test_external_has_tap variable somewhere before calling the first 502 test_external* function. See t9700-perl-git.sh for an example. 503 504 # The external test will outputs its own plan 505 test_external_has_tap=1 506 507 - test_external_without_stderr [<prereq>] <message> <external> <script> 508 509 Like test_external but fail if there's any output on stderr, 510 instead of checking the exit code. 511 512 test_external_without_stderr \ 513 'Perl API' \ 514 "$PERL_PATH" "$TEST_DIRECTORY"/t9700/test.pl 515 516 - test_expect_code <exit-code> <command> 517 518 Run a command and ensure that it exits with the given exit code. 519 For example: 520 521 test_expect_success 'Merge with d/f conflicts' ' 522 test_expect_code 1 git merge "merge msg" B master 523 ' 524 525 - test_must_fail <git-command> 526 527 Run a git command and ensure it fails in a controlled way. Use 528 this instead of "! <git-command>". When git-command dies due to a 529 segfault, test_must_fail diagnoses it as an error; "! <git-command>" 530 treats it as just another expected failure, which would let such a 531 bug go unnoticed. 532 533 - test_might_fail <git-command> 534 535 Similar to test_must_fail, but tolerate success, too. Use this 536 instead of "<git-command> || :" to catch failures due to segv. 537 538 - test_cmp <expected> <actual> 539 540 Check whether the content of the <actual> file matches the 541 <expected> file. This behaves like "cmp" but produces more 542 helpful output when the test is run with "-v" option. 543 544 - test_line_count (= | -lt | -ge | ...) <length> <file> 545 546 Check whether a file has the length it is expected to. 547 548 - test_path_is_file <path> [<diagnosis>] 549 test_path_is_dir <path> [<diagnosis>] 550 test_path_is_missing <path> [<diagnosis>] 551 552 Check if the named path is a file, if the named path is a 553 directory, or if the named path does not exist, respectively, 554 and fail otherwise, showing the <diagnosis> text. 555 556 - test_when_finished <script> 557 558 Prepend <script> to a list of commands to run to clean up 559 at the end of the current test. If some clean-up command 560 fails, the test will not pass. 561 562 Example: 563 564 test_expect_success 'branch pointing to non-commit' ' 565 git rev-parse HEAD^{tree} >.git/refs/heads/invalid && 566 test_when_finished "git update-ref -d refs/heads/invalid" && 567 ... 568 ' 569 570 - test_pause 571 572 This command is useful for writing and debugging tests and must be 573 removed before submitting. It halts the execution of the test and 574 spawns a shell in the trash directory. Exit the shell to continue 575 the test. Example: 576 577 test_expect_success 'test' ' 578 git do-something >actual && 579 test_pause && 580 test_cmp expected actual 581 ' 582 583Prerequisites 584------------- 585 586These are the prerequisites that the test library predefines with 587test_have_prereq. 588 589See the prereq argument to the test_* functions in the "Test harness 590library" section above and the "test_have_prereq" function for how to 591use these, and "test_set_prereq" for how to define your own. 592 593 - PERL & PYTHON 594 595 Git wasn't compiled with NO_PERL=YesPlease or 596 NO_PYTHON=YesPlease. Wrap any tests that need Perl or Python in 597 these. 598 599 - POSIXPERM 600 601 The filesystem supports POSIX style permission bits. 602 603 - BSLASHPSPEC 604 605 Backslashes in pathspec are not directory separators. This is not 606 set on Windows. See 6fd1106a for details. 607 608 - EXECKEEPSPID 609 610 The process retains the same pid across exec(2). See fb9a2bea for 611 details. 612 613 - SYMLINKS 614 615 The filesystem we're on supports symbolic links. E.g. a FAT 616 filesystem doesn't support these. See 704a3143 for details. 617 618 - SANITY 619 620 Test is not run by root user, and an attempt to write to an 621 unwritable file is expected to fail correctly. 622 623 - LIBPCRE 624 625 Git was compiled with USE_LIBPCRE=YesPlease. Wrap any tests 626 that use git-grep --perl-regexp or git-grep -P in these. 627 628 - CASE_INSENSITIVE_FS 629 630 Test is run on a case insensitive file system. 631 632Tips for Writing Tests 633---------------------- 634 635As with any programming projects, existing programs are the best 636source of the information. However, do _not_ emulate 637t0000-basic.sh when writing your tests. The test is special in 638that it tries to validate the very core of GIT. For example, it 639knows that there will be 256 subdirectories under .git/objects/, 640and it knows that the object ID of an empty tree is a certain 64140-byte string. This is deliberately done so in t0000-basic.sh 642because the things the very basic core test tries to achieve is 643to serve as a basis for people who are changing the GIT internal 644drastically. For these people, after making certain changes, 645not seeing failures from the basic test _is_ a failure. And 646such drastic changes to the core GIT that even changes these 647otherwise supposedly stable object IDs should be accompanied by 648an update to t0000-basic.sh. 649 650However, other tests that simply rely on basic parts of the core 651GIT working properly should not have that level of intimate 652knowledge of the core GIT internals. If all the test scripts 653hardcoded the object IDs like t0000-basic.sh does, that defeats 654the purpose of t0000-basic.sh, which is to isolate that level of 655validation in one place. Your test also ends up needing 656updating when such a change to the internal happens, so do _not_ 657do it and leave the low level of validation to t0000-basic.sh. 658 659Test coverage 660------------- 661 662You can use the coverage tests to find code paths that are not being 663used or properly exercised yet. 664 665To do that, run the coverage target at the top-level (not in the t/ 666directory): 667 668 make coverage 669 670That'll compile Git with GCC's coverage arguments, and generate a test 671report with gcov after the tests finish. Running the coverage tests 672can take a while, since running the tests in parallel is incompatible 673with GCC's coverage mode. 674 675After the tests have run you can generate a list of untested 676functions: 677 678 make coverage-untested-functions 679 680You can also generate a detailed per-file HTML report using the 681Devel::Cover module. To install it do: 682 683 # On Debian or Ubuntu: 684 sudo aptitude install libdevel-cover-perl 685 686 # From the CPAN with cpanminus 687 curl -L http://cpanmin.us | perl - --sudo --self-upgrade 688 cpanm --sudo Devel::Cover 689 690Then, at the top-level: 691 692 make cover_db_html 693 694That'll generate a detailed cover report in the "cover_db_html" 695directory, which you can then copy to a webserver, or inspect locally 696in a browser.