This library of functions are used by UNICOS tests to report results to standard out in a consistent manner. It is assumed that tests using this library have a distinct number of test cases, and that each test case is distinct and uniquely identified by the test case number. It is also assumed that test case results are printed in consecutive order, starting with 1. The library maintains a set of global variables (TCID, TST_TOTAL, Tst_count, Tst_range, Tst_lpstart, Tst_lptotal), which are used by the various functions to format the results and to keep track of the current result reporting state (i.e. how many total test cases there are, and how many have been reported so far) for the calling test.

The TCID and TST_TOTAL global variables are externed in the library, and MUST be defined/initialized by tests using the library. TCID must be set to the Test Case IDentifier, and TST_TOTAL must be set to the total number of test cases that will be reported.

The other global variables are available as externs to tests linked with tst_res.o. Tst_count is the running count of the number of test results that have been reported so far. The library initializes it to 0, and it should not be modified by the test. Tst_range, Tst_lpstart, and Tst_lptotal are used by to control how many test results are reported under certain conditions, by some of the functions in this library. The details are described below under the appropriate functions.

ttype

test result type; one of TPASS, TFAIL, TBROK, TCONF, TRETR, TWARN, or TINFO (explained below).

fname

pointer to a character string holding the name of a file whose contents will be printed on a new line following tmesg. If this pointer is NULL, it is ignored.

tmesg, [arg ...]

pointer to a character string containing a message explaining the test result. This character string can contain percent-sign conversion specifications as in printf(3C) with corresponding arg arguments following the tmesg argument.

NOTE: These routines use static space internally to hold the expanded message. The maximum size allowed for an expanded message is 2048 bytes.

func

pointer to a function which performs either the cleanup necessary prior to exiting the test or some function executed at the end of each iteration of a loop.

The possible test result types defined in test.h are as follows:

TPASS

The test case produced expected results.

TFAIL

The test case produced unexpected results.

TBROK

A resource needed to execute the test case was not available (e.g. a temporary file could not be opened).

TCONF

The test case was not appropriate for the current hardware or software configuration (e.g. MLS was not enabled).

TRETR

The test case was no longer valid and has been "retired."

TWARN

The testing procedure caused undesirable side effects that did not affect test results (e.g. a temporary file could not be removed after all test results were recorded).

TINFO

An informative message about the execution of the test that does not correspond to a test case result and does not indicate a problem.

tst_res() and tst_resm() are the basic result reporting functions. They report 1 or more test case results of the specified ttype. All result types are valid for these functions. The Tst_range global variable indicates the number of results that will be reported for each call to one of these functions. It is initialized by the library to 1, but may be set to a value > 1 by the test. Each call to one of these functions will result in Tst_range test case results being reported, each with an identical message (tmesg). Tst_range is always reset to 1 by the library before returning to the caller. If tst_res() is called with a fname argument, the contents of the file will only be printed for the first reported result. tst_res() takes the fname argument whereas tst_resm() does not.

NOTE: All calls to tst_res() specifying a ttype of TWARN or TINFO will be printed with a test case number of zero. Because of this, a Tst_range value > 1 is not valid for these types.

tst_brk() and tst_brkm() are used to report results for all test cases remaining in the test, and then call a cleanup function. The only result types that are valid for these functions are: TFAIL, TBROK, TCONF, and TRETR. When called with a ttype of TFAIL or TBROK, one result of the specified ttype will be printed, followed by results of type TBROK for the remaining test cases. When called with a ttype of TCONF or TRETR, the specified ttype will be printed for all remaining test cases. If func is not NULL, tst_brk() and tst_brkm() will call func after all results have been printed. If the call to func returns, tst_brk() and tst_brkm() will then call tst_exit(). If func is NULL, tst_brk() and tst_brkm() return to the caller after all results have been printed. If tst_brk() is called with a fname argument, the contents of the file will only be printed for the first reported result. tst_brk() takes the fname argument whereas tst_brkm() does not.

tst_brkloop() and tst_brkloopm() work just like tst_brk() and tst_brkm(), except they only report results for a the test cases remaining in the current loop. The Tst_lpstart and Tst_lptotal global variables are used to determine how many test cases to report results for. Prior to the start of the loop, the test must set Tst_lpstart to Tst_count, and Tst_lptotal to the number of test cases in the loop. If a test calls tst_brkloop() or tst_brkloopm() without first setting Tst_lpstart and Tst_lptotal to meaningful values, a WARN result will be reported and control will be immediately returned to the caller. This will most likely cause the result messages to be out of order. Unlike the tst_brk() functions, tst_brkloop() and tst_brkloopm() will not call tst_exit() if the func argument is not NULL and returns. tst_brkloop() takes the fname argument whereas tst_brkloopm() does not.

tst_flush() is used to print any results pending because of CONDENSE or NOPASS modes (described below), and flushes the output stream.

tst_exit() is used to allow tests to exit with a meaningful exit value. A bit is set in the exit value for each of the non passing test case result types (TFAIL, TBROK, and TWARN) encountered by the library. Thus any bit which is set in the exit value indicates that the corresponding result flag was used at least once in the test run.

The current bit fields for the result types are as follows:

TPASS

0000 /* .... .... */

TFAIL

0001 /* .... ...1 */

TBROK

0002 /* .... ..1. */

TWARN

0004 /* .... .1.. */

TRETR

0010 /* .... 1... */

TINFO

0020 /* ...1 .... */

TCONF

0040 /* ..1. .... */

NOTE: TPASS, TRETR, TINFO, and TCONF do not have an effect on the test program exit status.

tst_environ() is used to ensure that results reported by this library will go to the original stdout, even if the test changes the original stdout to another file, or closes it. A test may want to do this in order to redirect output that normally goes to stdout (e.g. printf() output) to a file. tst_environ() returns 0 upon successful completion, and -1 if it encountered any problems.

Four output display modes are supported by the tst_res() family of functions to enhance output readability. The active mode is controlled via the environment variable TOUTPUT, which must be set prior to the start of the test in order to have any effect (see ksh(1) for information on environment variables). The supported modes are as follows:

VERBOSE

A test result output line is generated for each test result. This is the default mode.

CONDENSE

Consecutive, identical PASS, FAIL, BROK, CONF, and RETR test results are condensed into one output line. The test case number field contains the range of results involved. WARN and INFO output lines are not condensed, but printed as usual.

NOPASS

All PASS, CONF, INFO, and RETR output lines are discarded (i.e. not printed), and consecutive, identical FAIL and BROK output lines are condensed as in CONDENSE mode. WARN output lines are printed as usual.

DISCARD

All output lines are discarded.