[Pdx-pm] sneaking tests into work
kevin at scaldeferri.com
Wed Feb 21 10:37:09 PST 2007
On Feb 21, 2007, at 8:17 AM, Michael G Schwern wrote:
> Ovid wrote:
>> --- Eric Wilhelm <scratchcomputing at gmail.com> wrote:
>>> If you're under a lot of time pressure, you might be told that you
>>> shouldn't be stopping to write tests. In this case, making them
>>> of your development should help.
>> When I first learned testing, I made the mistake of including the
>> for that as a separate line item on an estimate I turned in. My boss
>> scratched it off. Ever since then, I folded "time for testing" into
>> any estimate I turned in.
> The argument I like to use is that "non-productive" items like
> "commenting" or "indenting" or "version control" are not separate
> line items either. Why should testing be?
Our team considers it part of the code review process. Before a
review can even start, the author must provide a packet with the
code, the tests, coverage report for the tests, and (for C code)
valgrind output from the tests. Lack of test coverage and memory
leaks are grounds for preemptively failing the review.
(The fact that these policies have applied since the first line of
code was written does make it a lot easier to be this draconian.)
More information about the Pdx-pm-list