1To run the tests: 2 3 $ make check 4 5Note that if your /bin/sh doesn't support shell functions, you'll 6have to try something like this, where "/bin/sh5" is replaced by the 7pathname of a shell which handles normal shell functions: 8 9 $ make SHELL=/bin/sh5 check 10 11Also note that you must be logged in as a regular user, not root. 12 13WARNING: This test can take quite a while to run, esp. if your 14disks are slow or over-loaded. 15 16The tests work in /tmp/cvs-sanity (which the tests create) by default. 17If for some reason you want them to work in a different directory, you 18can set the TESTDIR environment variable to the desired location 19before running them. 20 21The tests use a number of tools (awk, expr, id, tr, etc.) that are not 22required for running CVS itself. In most cases, the standard vendor- 23supplied versions of these tools work just fine, but there are some 24exceptions -- expr in particular is heavily used and many vendor 25versions are deficient in one way or another. Note that some vendors 26provide multiple versions of tools (typically an ancient, traditional 27version and a new, standards-conforming version), so you may already 28have a usable version even if the default version isn't. If you don't 29have a suitable tool, you can probably get one from the GNU Project (see 30http://www.gnu.org). At this writting, expr and id are both part of the 31GNU shellutils package, tr is part of the GNU textutils package, and awk 32is part of the GNU gawk package. The test script tries to verify that 33the tools exist and are usable; if not, it tries to find the GNU 34versions and use them instead. If it can't find the GNU versions 35either, it will print an error message and, depending on the severity of 36the deficiency, it may exit. There are environment variables you can 37set to use a particular version of a tool -- see the test script 38(src/sanity.sh) for details. 39 40Some of the tests use fairly long command lines -- this usually isn't a 41problem, but if you have a very short command line length limit (or a 42lot of environment variables), you may run into trouble. Also, some of 43the tests expect your local timezone to be an integral number of hours 44from UTC -- if you usually use a fractional timezone, use a different 45(integral) timezone when running the tests to avoid spurious failures. 46 47If running the tests produces the output "FAIL:" followed by the name 48of the test that failed, then the details on the failure are in the 49file check.log. If it says "exit status is " followed by a number, 50then the exit status of the command under test was not what the test 51expected. If it says "** expected:" followed by a regular expression 52followed by "** got:" followed by some text, then the regular 53expression is the output which the test expected, and the text is the 54output which the command under test actually produced. In some cases 55you'll have to look closely to see how they differ. 56 57If output from "make remotecheck" is out of order compared to what is 58expected (for example, 59 60 a 61 b 62 cvs foo: this is a demo 63 64is expected and 65 66 a 67 cvs foo: this is a demo 68 b 69 70is output), this is probably a well-known bug in the CVS server 71(search for "out-of-order" in src/server.c for a comment explaining 72the cause). It is a real pain in running the testsuite, but if you 73are lucky and/or your machine is fast and/or lightly loaded, you won't 74run into it. Running the tests again might succeed if the first run 75failed in this manner. 76 77For more information on what goes in check.log, and how the tests are 78run in general, you'll have to read sanity.sh. Depending on just what 79you are looking for, and how familiar you are with the Bourne shell 80and regular expressions, it will range from relatively straightforward 81to obscure. 82 83If you choose to submit a bug report based on tests failing, be 84aware that, as with all bug reports, you may or may not get a 85response, and your odds might be better if you include enough 86information to reproduce the bug, an analysis of what is going 87wrong (if you have the time to provide one), etc. The check.log 88file is the first place to look. 89 90ABOUT STDOUT AND STDERR 91*********************** 92 93The sanity.sh test framework combines stdout and stderr and for tests 94to pass requires that output appear in the given order. Some people 95suggest that ordering between stdout and stderr should not be 96required, or to put it another way, that the out-of-order bug referred 97to above, and similar behaviors, should be considered features, or at 98least tolerable. The reasoning behind the current behavior is that 99having the output appear in a certain order is the correct behavior 100for users using CVS interactively--that users get confused if the 101order is unpredictable. 102 103ABOUT TEST FRAMEWORKS 104********************* 105 106People periodically suggest using dejagnu or some other test 107framework. A quick look at sanity.sh should make it clear that there 108are indeed reasons to be dissatisfied with the status quo. Ideally a 109replacement framework would achieve the following: 110 1111. Widely portable, including to a wide variety of unices, NT, Win95, 112OS/2, VMS, probably DOS and Win3, etc. 113 1142. Nicely match extended regular expressions of unlimited length. 115 1163. Be freely redistributable, and if possible already the kind of 117thing people might have already installed. The harder it is to get 118and install the framework, the less people will run the tests. 119 120The various contenders are: 121 122* Bourne shell and GNU expr (the status quo). Falls short on #1 123(we've only tried unix and NT, although MKS might help with other DOS 124mutants). #3 is pretty good (the main dependency is GNU expr which is 125fairly widely available). 126 127* Bourne shell with a new regexp matcher we would distribute with 128CVS. This means maintaining a regexp matcher and the makefiles which 129go with it. Not clearly a win over Bourne shell and GNU expr. 130 131* Bourne shell, and use sed to remove variable portions of output, and 132thus produce a form that can be compared with cmp or diff (this 133sidesteps the need for a full regular expression matcher as mentioned 134in #2 above). The C News tests are said to work this way. This would 135appear to rely on variable portions of output having a certain syntax 136and might spuriously recognize them out of context (this issue needs 137more investigation; it isn't clear how big a problem it is in 138practice). Same portability issues as the other choices based on the 139Bourne shell. 140 141* Dejagnu. This is overkill; most of dejagnu is either unnecessary 142(e.g. libraries for communicating with target boards) or undesirable 143(e.g. the code which stats every file in sight to find the tests). On 144the plus side, dejagnu is probably closer than any of the other 145choices to having everything which is needed already there. 146 147* Write our own small framework directly in tcl and distribute with 148CVS. The tests would look much like dejagnu tests, but we'd avoid the 149unnecessary baggage. The only dependency would be on tcl (that is, 150wish). 151 152* perl or python or <any other serious contenders here?> 153 154It is worth thinking about how to: 155 156a. include spaces in arguments which we pass to the program under 157test (sanity.sh dotest cannot do this; see test rcs-9 for a 158workaround). 159 160b. pass stdin to the program under test (sanity.sh, again, handles 161this by bypassing dotest). 162 163c. have a send-expect type dialog with the program under test 164 (e.g. see server-7 or pserver-4 which want to talk the CVS 165 protocol, or the many tests which need to answer the prompt of "cvs 166 release", e.g. deep-5). 167 168ABOUT ADDING YOUR OWN TESTS 169*************************** 170 171As stated in the HACKING file, patches are not accepted without documentation 172and tests. Many people seem to be scared off by the large size of the 173sanity.sh script, but it is not really very complicated. 174 175You can probably ignore most of the begining of the script. This section 176just sets some environment variables and finds the tools the script needs to 177run. 178 179There is one main loop you can find by grepping for "The big loop". This loop 180repeatedly calls a case statement where the individual cases are of the form: 181 182 testname) 183 ... 184 ;; 185 186If you add a complete new test be sure to add it into the default list of tests 187(grep for 'tests=' near the begining of the script) as well as the case 188statement. During debugging, be aware that the sanity.sh usage allows for a '-f 189testname' option to continue through the default list "from" a particular test 190as well as interpreting everything in argv past the required options as test 191names to run individual tests. 192 193Within each major test section, individual tests usually look like: 194 195 dotest testname-subtestname "shell command" "optionally multiline regexp" 196 197Tests should always start in $testdir and create a subdirectory to operate in 198and remove their cruft and end back in $testdir. The dotest functions output 199failure messages and exit if the shell command exits with the wrong exit code or 200its stdin/stderr output doesn't match the regexp. There are a few dotest 201variations, most notably dotest_fail for expected non-zero exit codes. 202 203Other than that the script is mostly vanilla Bourne shell. There are a few 204constructs used for versatility and portability. You can grep for the ones I 205miss, but here are a few important ones. I'm leaving off long explanations 206after the first few since it probably gives you the idea and the data is in 207sanity.sh. 208 209Note that the boolean variables contain shell commands which return true or 210false when executed and are intended to be used like, 211"if $remote; then ... ; else ... ; fi" 212 213 214 * $testdir = the directory this test is taking place in 215 (CVSROOT=$testdir/cvsroot or 216 CVSROOT=:fork:$testdir/cvsroot) 217 * $testcvs = full path to the cvs executable we are testing 218 * $PLUS = expr dependant uninterpreted '+' since this can vary 219 * $DOTSTAR = expr dependant _interpreted_ .* since some exprs don't 220 match EOL 221 * $username = the username of the user running the tests 222 * $username8 = the first 8 characters of $username, output by some 223 system and CVS commands 224 * $anyusername = regexp to match any valid system or CVS username 225 * $hostname = regexp to match a hostname 226 * $PROG = regexp to match progname in CVS error messages 227 * $remote = ':' (true) or 'false', depending on whether the script is 228 running with a remote CVSROOT 229 * $keep = ':' (true) or 'false'. When set, the first test run will 230 leave any files and directories it created in $testdir and 231 exit when complete. 232 233And, of course, some characters like '.' in regexps need to be '\' escaped when 234you mean them literally. Some characters may be interpreted by the shell, 235e.g. backquotes and '$', are usually either escaped or replaced with '.'. 236dotest adds the final '$' anchor to the regexp itself and all the expr 237implementations I know of implicitly supply the start anchor ('^'). 238 239If you only make a few mistakes, the work is, of course, still usable, though we 240may send the patch back to you for repair. :) 241