home *** CD-ROM | disk | FTP | other *** search
- Newsgroups: comp.object,comp.lang.c++,comp.realtime,comp.dcom.telecom.tech,comp.arch.embedded
- Path: vtnt02.vantek.net!not-for-mail
- From: Sid Johnson <sjohnson@mail.vantek.net>
- Subject: Re: Can OO be successful in real-time embedded systems?
- Message-ID: <316F437B.40A0@mail.vantek.net>
- Date: Fri, 12 Apr 1996 23:02:35 -0700
- X-Mailer: Mozilla 2.0 (Win16; I)
- MIME-Version: 1.0
- References: <316BF0C5.1FE1@condat.de>
- Content-Type: text/plain; charset=iso-8859-1
- Content-Transfer-Encoding: 8bit
-
- Henning Rietz wrote:
- >
-
- >
- > - "OO systems are too slow"
- > - "OO systems eat up too much memory"
- >
- > I believe, that this is not necessarily true but heavily related to
- > experience. Now I┤m asking you:
- >
- > How far "down" does the application of OO really go?
- > How far will it go in the future?
- > Who develops commercial(!) embedded real-time systems
- > using OO methods and languages?
- > Will OO ever be of major importance in that area?
- >
- >
- >
- >The trick is to remember that C++ is just doing what you would like to do in C, but
- doing it with better looking source code. It provides modularity and extensibility without
- cluttering source code with pointers and it avoids global data without the call overhead of "get"
- functions (inline fns). These are great maintainability advantages.
-
- As for being a resource hog, that is a matter of self-discipline. If you avoid exceptions
- and templates, are reasonable with inheritance, and hit a happy medium on what you call an
- object, there is very little overhead. In addition, for "single-instance" objects, you can make
- the data and fns static and avoid the "this" pointer. Even with virtual fns, if the objects are
- named at compile time instead of run-time, the virtual table is avoided altogether.
-
- We are currently reengineering two existing products (similar but different) into one product
- line using C++. These products are characterized by several choices of optional peripherals, and
- very complex interactions of internal features. We have found C++ to be just the tool for
- creating an architecture which simplifies the view of this software and gives the flexibility to
- mix and match peripherals.
-
- As for tools, there seems to be nothing below 32-bit except in the x86 family. This is because
- the Microsoft and Borland compilers support it. Several other vendors provide linkers, locaters,
- and debuggers to support these compilers. With this approach, you can go as low as the x86
- family goes. As for OOA/D, the Rose tool supports these compilers.
-
- I think the industry is really missing an opportunity to turn out much better software without
- serious performance degradation -- even in the 8/16-bit market. :-)
-
-
- --
- __________________________________________________________________
- Sid sjohnson@vantek.net
-
- Come visit @ the Philosopher's Corner - http://www.vantek.net/pages/sjohnson/
-
- In thinking, keep to the simple. In conflict, be fair and generous.
- In governing, don't try to control. In work do what you enjoy.
- In family life, be completely present.
- Lao-Tzu
- ___________________________________________________________________
-
-
-