Only one bone of contention here:
Depends on the coverage (sorry BrowserUK, this *does* matter). It is easy to make a module to pass all tests if there is only one test that tests nothing: ok (1, "Make it PASS!"); is a legal but useless test
Coverage tools have there place; and that place is for authors, not users!
An automated coverage tool can act as a sanity check; as a guide for the author to decide if s/he has written all the test s/he believes are necessary.
But the moment the numerical summation produced by a dumb code evaluator becomes a tool for users to judge the quality of an author's work and decisions, software development goes to hell in a hand basket.
Until code coverage tools get smart enough to actually write the tests for all the places they suggest need one; they should serve only as suggestions to the human being charged with writing them. And any user who believes that 100% coverage tells them anything about the quality of the testing is in for a very rude awakening.
In the end, always run my own sanity check of the functionality I import from 3rd party modules, and if it appears to produce the correct results for my usage, all the gibberish numbers (supposedly good or bad) produced by the automated test/coverage/kwality/et al tools and processes are entirely meaningless.
With the rise and rise of 'Social' network sites: 'Computers are making people easier to use everyday'
Examine what is said, not who speaks -- Silence betokens consent -- Love the truth but pardon error.
In the absence of evidence, opinion is indistinguishable from prejudice.
Suck that fhit
|