Subject:
|
Re: Interests in non-GNU C compiler for RIS?
|
Newsgroups:
|
lugnet.robotics.rcx
|
Date:
|
Tue, 26 Oct 1999 22:35:48 GMT
|
Viewed:
|
1375 times
|
| |
|
|
In lugnet.robotics.rcx, Richard F. Man writes:
> I know most lego alternative SW seems to be Open Source, but please
> don't flame me for asking this question - ImageCraft is a company that
> provides low cost C compilers for 8/16 bit microcontrollers such as
> Atmel AVR, Moto HC11/12/16. While not Free as in Open Source, or Free as
> in beer, we pride ourselves on providing easy to use, professional
> quality tools at price 10x lower than that of traditional embedded
> compilers. And robotics is my first love - in fact, the company was
> started when I try to productize my subsumption architecture based
> multitasking executive and found out that there is no low cost C
> compiler out there. In fact, there are some mentioning of our ICC11
> tools in the handyboard newsgroup.
>
> Anyway, all this is a long-winded introduction to ask if anyone is
> interested in having an easy to use ANSI C compiler for their RIS. I
> know GCC is available and there is NQC, which looks to be a fantastic
> piece of work, and of course there are pbForth and the JavaVM port in
> progress. What we will bring to the community is an alternative to these
> tools is an easy to use Windows environment, plus full C support (4 byte
> longs and 4 bytes floats). If we do such a thing, we will make it
> interoperable with LegOS etc. of course. We even sell Linux version of
> our command line tools, FWIW.
>
> Please let me know what you think. The pricing on our current product is
> $160-$199, but if there is a larger market for this, we can consider
> dropping it to $50-$100 range. Our goal is to make money - this is our
> day (and night and weekend) job, but if we do it, you can be sure that
> it will be a very usable environment to program the RIS in - since I
> will be one of the biggest users :-)
>
> Another interesting possibility is that I do not believe execution speed
> is of primary concern, so instead of doing a native C compiler, may be
> we can release a Virtual Machine based compiler. We can release the API
> of the VM and the VM itself as Open Source for people to hack around
> too. Doing this is also easier than doing a native compiler, so perhaps
> we can release this faster.
> --
> // richard
> http://www.imagecraft.com
I personally would love to see a development system for the RCX that produces
native H8 machine code, and that doesn't require a huge investment in time and
disk space to install and configure.
As I see it, there are already many development systems out there for free
that are this simple to install and use, as long as you can live with the
limitations of the RCX firmware. What your company could bring to the table
would be a small toolset that can create highly optimized code to run on the
RCX. (My bias against loading a VM onto the poor little RCX is showing
through here...) As I understand it, the standard RCX firmware takes several
milliseconds to interpret each byte-code opcode. This makes any "thinking"
operations slow, and also prevents the RCX from being used for any control
application that requires more than about a 10 Hertz control loop. Compiling
C to native H8 code would take a huge burden off of the H8 at runtime, and
transfer that burden to my desktop machine at compile time. This would be a
HUGE improvement over any alternative programming system that I am aware of
except for LegOS. And no offense to the LegOS crowd, but it has the
reputation for being notoriously cranky to install, which is really the only
thing that keeps me away.
It would also be absolutely fantastic to be able to do floating-point
calculations on the RCX when necessary. (I've got a few applications that
could benefit from a little basic trigonometry.) I imagine your system would
also support dynamic memory allocation, arrays, etc.
One feature that is supported by all of the RCX programming systems that I've
used is multi-tasking in one form or another. I've found it difficult to
actually use simultaneous tasks to do very much (due to the lack of semaphore
support in the native firmware), but you will probably find that your intended
customer base will expect this feature.
As far as price goes, well, obviously cheaper is better. You will be
competing against a variety of established, FREE environments, so even if your
product is 1000 times better, you're probably not going to sell a lot of
copies if the price is more than $100. I would suspect that you would sell
ten times as many at $50 a copy. After all, the RCX is a *toy*, and many
people who use it have a hard time justifying a huge outlay of cash for a
hobby. This is one area where your business model will be significantly
different from the traditional customer base for embedded system compilers.
But you should also be able to achieve a much higher volume, since the RCX is
a consumer channel product and you will be selling the development tools
directly to end users. (I don't know how many RCXes have been sold, but it's
gotta be pushing 200,000 units worldwide by now...)
Sign me up! I'll buy a copy when it's ready. (Heck, let me know if you need
beta testers!)
|
|
Message is in Reply To:
 | | Interests in non-GNU C compiler for RIS?
|
| I know most lego alternative SW seems to be Open Source, but please don't flame me for asking this question - ImageCraft is a company that provides low cost C compilers for 8/16 bit microcontrollers such as Atmel AVR, Moto HC11/12/16. While not Free (...) (25 years ago, 26-Oct-99, to lugnet.robotics.rcx)
|
4 Messages in This Thread:
 
 
- Entire Thread on One Page:
- Nested:
All | Brief | Compact | Dots
Linear:
All | Brief | Compact
|
|
|
|
