>>>>> "NT" == Nathan Torkington <[EMAIL PROTECTED]> writes:
NT> Once we have an API, we write test cases to exercise the API.
NT> Then we code.
Would it be worthwhile to 'waste' some time on writing caller and callee
stubs? This would flesh out the final api and give the module writers
something of an environment to work with?
I'd actually include, 'random' failures into the stubs to give a chance
to excersize the error legs.
On one project that I worked on, I put the failure generation in the
malloc routine. It failed one call later than the last successful
call. (Global tied in with the main driver code.) This find a _huge_
number of bugs in my error handling.
<chaim>
--
Chaim Frenkel Nonlinear Knowledge, Inc.
[EMAIL PROTECTED] +1-718-236-0183
