home *** CD-ROM | disk | FTP | other *** search
- Path: news.Stanford.EDU!not-for-mail
- From: bims@isl.Stanford.EDU (Harry V. Bims)
- Newsgroups: comp.object,comp.lang.c++,comp.realtime,comp.dcom.telecom.tech,comp.arch.embedded
- Subject: Re: Can OO be successful in real-time embedded systems?
- Date: 19 Apr 1996 11:40:13 -0700
- Organization: Stanford University
- Message-ID: <4l8mmd$s5o@isl.Stanford.EDU>
- References: <316BF0C5.1FE1@condat.de> <dibaldDpnpBH.5Et@netcom.com>
- NNTP-Posting-Host: isl.stanford.edu
-
- In article <dibaldDpnpBH.5Et@netcom.com>,
- Dave Baldwin <dibald@netcom.com> wrote:
- >Henning Rietz (rietz@condat.de) wrote:
- >
- >: I can say "everybody" is using OO in some areas (mainly network
- >: management, switch provisioning, customer care), BUT there are (almost)
- >: no examples in the area of (small) embedded systems, main reasons for
- >: that being:
- >
- >: - "OO systems are too slow"
- >: - "OO systems eat up too much memory"
-
- Quite the contrary. Here are Wireless Access, I have designed and built
- a real-time, objected-oriented system running on a PC. The system
- creates a local area two-way paging environment. It schedules and
- manages multiple channels and users simultaneously. A coomparable system
- was implemented without OO technology, and it requires a 250MHz Alpha to
- do the same job. In addition, code changes can occur much more quickly
- when you have an OO framework to start with. The problem is that
- very few people know how to create good OO designs. It takes quite
- a long time to climb the learning curve.
-
- As far as memory is concerned, there is some overhead associated with
- C++ compilation versus C, however, once your program reaches in excess
- of 15K lines, that no longer becomes an issue. In fact, because of code
- and data reuse, my code is more efficient from a memory requirements
- perspective than the non-OO counterpart.
-
- >
- >Object-dis-oriented programming is (like some others) intended to hide
- >the hardware from the programmer. How useful can this possibly be when
- >small embedded systems are expressly for dealing with the hardware? Some
- >of the techniques can be useful, but the overhead and 'hiding' is exactly
- >what you don't need in hardware control.
-
- Hardware designs are subject to change. When that happens, it is much
- more difficult to adapt your code when you DON'T use OO. Again, overhead can
- be minimized through careful design.
-
- >There is no universal programming method. Even the examples you cite are
-
- You are right. However, if you are building a program that is longer than
- about 10K lines, OO is generally the best methodology.
-
- Harry Bims
- Senior MTS
- Wireless Access, Inc.
- 408-653-2288
- disciple@waccess.com
-
-