home *** CD-ROM | disk | FTP | other *** search
- Path: sparky!uunet!spool.mu.edu!olivea!apple!applelink.apple.com
- From: PHILMONT@AppleLink.Apple.COM (Philmont SW Mill, Jesse Feiler,PAS)
- Newsgroups: comp.sys.mac.oop.macapp3
- Subject: Re: Takeover - QA
- Message-ID: <726156743.3640369@AppleLink.Apple.COM>
- Date: 4 Jan 93 14:05:00 GMT
- Sender: daemon@Apple.COM
- Organization: AppleLink Gateway
- Lines: 37
-
- Dave,
-
- I beg to differ with your approach to QA. Bernadette is right in that a full
- set of QA is needed for every bug fix (or set thereof).
-
- You discuss a bug-fix and use phrases like "guaranteed" and "cannot possibly
- cause..." Don't tempt fate.
-
- Without discussing this particular bug, I can easily conjure up a number of
- scenarios where this type of bug fix could cause problems. E.g., pushing the
- size of a code segment up by a few bytes - just enough to cause overflows in an
- application that links user code into that segment.
-
- I participated in a "harmless bug fix" that inserted an extra blank in a title
- string in order to correct a slight off-centering. Because the gods of software
- were playful that day, this insertion of ONE BLANK (" ") in the format string
- led directly to the total collapse of the major financial data entry system at
- a very sensitive institution. Data entry workers who should have started keying
- at 8 AM started at 2:30 PM.
-
- I periodicially look at the official bug list, and -- as most people do -- I
- file away MacApp.Tech$ links that look as if they may touch on bugs/solutions
- that we may need in the future. We apply bug fixes if we need them (and we
- usually can do them in subclasses of our own) and rely on MacApp which has been
- thoroughly QA'd to remain imperfect but stable.
-
- Finally, you write, "[why do all of these discussions insist that all bug work
- and QA needs to be done in a vacuum???]". The answer is because the greater the
- degree of separation between the coder and the tester, the more likely the
- testing is to be thorough. I know of one case in which the testing staff isn't
- even told what specific bug fixes are to be tested: they are to insure the
- complete integrity of the new code; on a second pass, they test the specific
- fixes.
-
- Jesse
-
-
-