home *** CD-ROM | disk | FTP | other *** search
- Here are a basic set of confidence tests for the C library: a lot of
- work is needed to build up this set of tests and i seek your active
- participation and contributions. The principles i've tried to follow
- are:
- - the tests should be simple. larger tests go into the sub-dir
- big/. Performance type tests can go into perf/. The tests mostly
- test functionality. Of course this is not enough, and we should
- have performance tests alongside the functional tests, to flag
- when we break things big time performance wise (both in time and
- space). We also need to develope coverage tests.
-
- - the tests should be self contained: you should not have to examine
- the output, or diff the output against some file to figure out
- if the output is indeed correct, the test should check its own
- result.
-
- - tests should only print out stuff when something goes wrong,
- otherwise remain silent. Of course this is not always possible
- (or followed, see tstdio for example). in such cases a .out file
- exists for you to diff against.
-
- - all tests should return a valid status.
-
- - all the tests should be 16/32 bit integer clean.
-
- - a test should test a small number of related functionality. that
- way its easier to isolate stuff when things go wrong.
-
- naming:
- tests are named as tXXXXX.c, any input files for the test
- are named tXXXXX.in, and the executables are named tXXXXXX16.ttp
- and tXXXXX32.ttp for 16/32 bit ints respectively.
-
- cheers,
- --
- bang: uunet!cadence!bammi jwahar r. bammi
- domain: bammi@cadence.com
- GEnie: J.Bammi
- CIS: 71515,155
-