barriost@gtephx.UUCP (Tim Barrios) (04/10/91)
i wanted to find out what other companies' experience have been with
the benefits of doing code reviews/inspections before vs after unit
testing. i recently have read a couple of articles and
have been involved in some discussions with people on the topic.
at my company, we have historically developed huge, monolithic embedded
systems that could only be tested on a very costly prototype hardware
platform. as such, we had historically done code reviews before any
testing since testing was so costly. as such, code reviews included a
bit of 'playing the computer' to speculate as to whether the code
would work or not.
now, we have an improved tools environment for our large systems which
allows us to do unit testing on the engineer's workstation. also,
we are beginning to do projects using commercial languages (as opposed
to our own) which we can also use to run some level of unit testing
on the development platform instead of the target hardware. so,
projects have sometimes reversed the ordering and done code reviews
after the code had been unit testing.
the approach of doing unit testing before code reviews seems to make a
lot of sense in that it is a lot cheaper to let a computer work out
algorithm/logic problems. however, i have read/heard that the
reviewers then tend to not take the code review too seriously since
'it already works'. if so, this is very unfortunate, i think, since
such post-unit-testing reviews could concentrate on higher-level
issues like coupling, cohesion, maintainability, reusability, etc.
--
Tim Barrios, AG Communication Systems, Phoenix, AZ
UUCP: ...!{ncar!noao!asuvax | uunet!zardoz!hrc | att}!gtephx!barriost
Internet: gtephx!barriost@asuvax.eas.asu.edu
voice: (602) 582-7101 fax: (602) 581-4022