home *** CD-ROM | disk | FTP | other *** search
- Path: cliffy.lfwc.lockheed.com!news
- From: Ken Garlington <GarlingtonKE@lfwc.lockheed.com>
- Newsgroups: comp.lang.ada,comp.lang.c,comp.lang.c++
- Subject: Re: C/C++ crap knocks Ada
- Date: Thu, 07 Mar 1996 13:23:24 +0000
- Organization: Lockheed Martin Tactical Aircraft Systems
- Message-ID: <313EE34C.1DFB@lfwc.lockheed.com>
- References: <JSA.96Feb16135027@organon.com> <4hakfl$ogd@fred.netinfo.com.au> <313C749D.2C34@lfwc.lockheed.com> <4hk3qg$24ci@info4.rus.uni-stuttgart.de> <4hl106INNgjl@keats.ugrad.cs.ubc.ca>
- NNTP-Posting-Host: cliffy.lfwc.lockheed.com
- Mime-Version: 1.0
- Content-Type: text/plain; charset=us-ascii
- Content-Transfer-Encoding: 7bit
- X-Mailer: Mozilla 2.0 (Macintosh; I; 68K)
-
- Kazimir Kylheku wrote:
- >
- > Under conditions where ultra-high reliability is required, I'd add a lot of
- > redundant check bits to every word of storage, and let the hardware do much of
- > the error detection and correction. It would be extremely irresponsible to use
- > unreliable hardware where the requirement is for ultra-high reliability, no?
-
- Unfortunately, as you start adding complexity to the hardware, you start hitting
- other brick walls. Many of these systems, particularly the fail-operate ones,
- have to operate in extreme environmental conditions and various power failure modes.
- This limits the hardware complexity. Also, as you add complexity, the MTBF falls.
- Finally, as you add complexity, the chance of a hardware bug escalates. So, letting
- the hardware handle errors is not always the right answer.
-