files


NAME - tester files

tester files - tester documentation on files


DESCRIPTION

This page provides a brief explanation of the files used by tester. It discusses the affect of options. It explains the conventions used in names.

The tester manual overview refers to all other sections:


NOTES

* Files used by the tester
In looking at files two system defined names are used often. Check out $TESTSPEC and FILESPEC. The represent the more general $TESTNAME and $FILENAME names. These names are used to form other names as shown below.

stderr - Standard Error Output
The use of a stderr keyword insures that standard error output is captured to: $TESTSPEC .se

Standard error output is initially generated to $TESTSPEC .R_SE.

stdout - Standard Out Output
The use of a stdout keyword insures that Standard out output is captured to: $TESTSPEC .so

Standard out output is initially generated to to $TESTSPEC .R_SO.

both
The use of a keyword both will capture standard output and standard error output to $TESTSPEC .se and $TESTSPEC .so.

Both outputs are initially generated to $TESTSPEC .R_SE and $TESTSPEC .R_SO.

same
The use of the same keyword captures both standard out and standard error output to the same file, $TESTSPEC .seso.

The combined outputs are initially generated to $TESTSPEC .R_SESO.

In the cshell the order of output when combining standard out with standard error varies. So often a tester will just save combined output rather than expect it to be the same each time.

user tags
User tags are captured to $TESTSPEC .tag

User tags are generated to $TESTSPEC .R_TAG

expected results
Result files are captured to $TESTSPEC .result, $TESTSPEC .R_* and $TESTSPEC .S_*

Any FAILURES captured to $FILESPEC .result, $TESTSPEC .FAILED

Use the -generate option to re-generate results. By default the first run will generate data.

The resulting success/failure of all tests are appended to $TESTSPEC .result. The -summary option requests this file to be deleted.

After a test has run, if it was successful then only the expected output and a summary of results is kept around. The save command may be used to save additional output. The -keep options will keep those files around so they may be looked at.

Any FAILED test will leave a copy of all files to inspect the expected results and check out the failure.

.

* Name Substitution from a file
File name substitution is applied at test build time. Name inclusion is additive -- names stay defined until redefined.

.names
If the file .names exists, then it will be read for global names that can span all tests. This file of names is read first and local names will override names that are redefined.

other option using file
If the option -other is used when invoking the tester, an alternate test file can be applied to a test specification.

Any outside name file by default will override the local names specified in the test specification. This means that locally defined names as well as the .names file are not read.

other options with @file
-other @ will make this request additive. So the names file will be read first. Any local names will still be created and override names that are redefined. Once again, the .names file in this case will not be read.

.

* Test Ordering from database files
By default tests are ordered sequentially and saved to a database of test descriptions. The test description occurs after the test keyword.

Test Ordering
Test descriptions need to remain unique or they will write over each other.

This database should be saved with the results of the test so that other tests can be added without affecting the original sequential ordering of the tests.

The -without option turns off database ordering and forces simple sequential test ordering.

db.FILENAME.pag
This file is based on the name of the test specification, $FILENAME . This file contains a list indexed records. Each index is a test description. The record is the sequential index of each test. Tester adds one to this number to produce a test number.

This file should be saved with the results of the test so that other tests can be added without affecting the original sequential ordering of the tests.

db.FILENAME.dir
This file is based on the name of the test specification, $FILENAME .

This file contains a hash table into the indexed records discussed above.

This file should be saved with the results of the test so that other tests can be added without affecting the original sequential ordering of the tests.

db.FILENAME.lock
This file need not be saved. It is used by the tester as it makes updates to the test description database from multiple test runs.

.

1