We are running a survey to help us improve the experience for all of our members. If you see the survey appear, please take the time to tell us about your experience if you can.
I,ve installed the Compiler and I can,t get even the simplest code to compile properely.
Anyone know where the fix for this bug is?
Or is it a limit of the demonstration version?
void main(void) { cout << "Hello world!"; }
Someone in those companies must have seen an opportunity to fill the need for C++ on the 8051.
I think you don't understand what companies do.
Companies don't exist to fill needs.
Companies do exist to make money, by making products that there is a market for. It does not matter whether the product is necessary, useful or sensible. The only thing that matters is that there are people who will pay for it (or that can be talked into paying for it). In fact, there are many, many necessary, useful and sensible products that do not get made because there aren't enough people that can pay for them.
Maybe, just maybe, for some projects it would be a quick way of getting the job done.
If you had read this thread, then you would have already seen a mention of the one scenario where such a compiler makes sense - when there is an existing, large code base that needs to be run on an 8051 without porting it to a more appropriate language.
Also, you could already be halfway done with your project if you had taken advice from experienced embedded developers, instead of arguing with them. Or at least you could have read the specs of your hardware. Simply using cout and expecting text to magically appear on your display makes me believe that you expect your 8051 to work like a PC. It doesn't.
Who knows, it may even have people on that one who have slightly less narrow minded views.
"I know only C++, so any problem must be solved in C++" isn't narrow-minded ?
All the other people on this thread also know C++. They know C. They know assembly. They know the hardware architecture of the 8051. They have a broad perspective and know that and why C++ is not the right tool for simple 8-bit devices.
Anyway, good luck at what you're trying to do. You'll need it.
You seem to think that C++ with virtual methods is good. The 8051 chips hates function pointers, and a virtual method is a form of a function ponter.
You seem to think that object-orientation with dynamically created objects is good. The 8051 has too little memory to work with dynamic memory. Even if you can implement it, you run a large risk of failing because of memory fragmentation. So forget new/delete.
You probably haven't realized that the '51 chips are TRUE 8-bit chips. They are not 16-bit chips with an 8-bit bus like the 8088 was on the original PC, more than 20 years ago. The register size is smaller than the short or int data types, so you don't have atomic add/sub/mul/div. How much code do you think a software div contains?
The processor doesn't have a file system, so forget about file streams.
The target doesn't have an OS, and the processor are just minimally able to host a minimal OS. So forget about threads or tasks.
There is no driver layer. Every single hardware peripherial needs bit-fiddling code to function. Code written by you. Or written and uploaded by narrow minded old fools who doesn't know that C++ has come to town.
The processor contains true boolean variables. But the variables are not actually boolean true/false, but actually on/off.
The processor requires special language extensions for tagging of variables to inform in which memory areas they should be stored.
The application has to settle for using a subset of the C runtime library. A huge number of the C library doesn't make sens. A even huger part of the C++ standard libraries makes sens - or can even be used - on a '51 platform.
You have no experience of embedded development, but instead of listening, you think everyone is narrow-minded.
You are the drunk who stands on a rooftop and claims you can fly.
The problem is that you claim that you know C++, but your posts gives very clear indications that you do not actually know so much about C++. You do not seem to know what makes C++ ticks - what it looks like under the hood. For embedded programming on this scale, you have to know the inner workings. Programming on a macro scale does not work.
So you get a C++ compiler. You then have to use the C++ compiler to write a mostly C-ish program, because most of the C++ concepts and libraries are not available. Who do you expect to fool with that?
You have still totally missed the point. The people who are on this forum have already got things done. We are not arguing based on "maybe" or "i like to" or "it would be good to". We are arguing based on existing products available all around you. Real, physical, working factory-produced products.
"Companies don't exist to fill needs."
What a cynical view of life!
Ok, Mr Henry Ford wanted to make money but part of the dream was to give transport to the masses.
Part of marketing is understanding what people want!
I want to use C++ because I know C++.
I say that in the same way that I said I want a response in English. It is because I talk English!
To do something you are familiar with is quicker than doing something totally foreign.
To think knowledge in one area of life is directly applicable to other areas of life is not faster. It's just plain dumb!
Not knowing the relationship between C and C++ questions your C++ knowledge.
To think companies focuses on customer needs instead of what makes money is naive.
Have you picked up the phone and called IAR? Do they recommend C or C++ for a '51 target? Have you asked how much of the C++ concepts and libraries you can actually use?
Have you prepared critical questions to ask them, or do you make your neck as long as possible and tells your suppliers: please put a noose around my neck - I'm gullible?
This isn't about what you know - but what you don't know. If you focus your life around what you know, you will never grow. Right now you are a man with a hammer, desperately seeing everything as a nail.
Please define narrowminded...
I said PART of marketing I said PART of the dream
Surely you would agree that it is always better to build upon what you know rather than always going for something different.
A skyscraper is built of same style blocks one upon another. If they were all different then the structure almost certainly wouldn't stand!
I do not need to define narrow minded - Just use your favourite search engine.
In this instance, dismissing C++ out of academic principal I would say is narrow minded.
Of course. I would even say that all of marketing is understanding what people will want to buy, or what they can be made want to buy. It is all about business - if people will pay for it, then is is economically sensible for a company to make.
That, of course, does not mean at all that the product has to be technically sensible. What the customer does with the product does not concern the company, as long as they get paid.
We've already discussed this: If you know C++, then you also know C (if you do not know C, please stop claiming that you know C++).
I am sorry, but the hardware you have to work with isn't going to accomodate your wishes. Humans have this capability called "learning", while your hardware can only be used as-is, or replaced completely.
The concepts of C++ are totally foreign to your 8051. That is precisely why is does not make sense to try programming it in C++.
Hello ? The other posters on this thread are experienced embedded developers. That means they have already written code for real-life projects that are being sold on the market. Their views and opinions stem from years of practical, hands-on experience with the 8051 architecture.
The person who has a purely academic point of view is you.
sir filip
you be needing project work soon?
i help you please
i say send info and i see what to do for work
if working c then you think be happy yes??
I know what it means. I'm just very curious to know what you think it means.
It is an academcial goal to use as high-level language as possible for all development. It a practical rule to not use a higher-level tool than the target can handle in a good way. You call our arguments academical???
Many of us _are_ experienced C++ developers, which is something you seems to constantly ignore.
Don't you wonder just a little bit why I claim to be an experienced C++ developer, and still says that C is a better language to use for a '51 chip?
Don't you get a feeling that there _may_ be parts of the equation that you haven't seen yet?
How long was it since you left school?
How many real projects (embedded or not) have you worked with?
Haven't you noticed that real projects tend to have a large number of mutually exclusive requests - something that didn't exist in school assignments. In real projects, you always have to compromise!
"I do not need to define narrow minded - Just use your favourite search engine"
Just tried doing a search for that on my favourite search engine and got nothing!?
Trouble is, my favourite search engine is http://www.booble.com ;)
Maybe you should have been more specific.
I noticed the following comment upthread:
As I have said before: The problem you are going to run into if you use the Ceibo + Keil combination is that the evaluation version of Keil C51 only allows a code size of 4 kB. Any complex string formatting function (this includes printf and cout) will easily need upwards of 10 kB code space.
Compiling the following program with a recent version of C51:
#include <stdio.h> void main(void) { printf("Hello world\n"); while(1); }
Gave the following map:
BASE START END USED MEMORY CLASS ========================================================== X:000000H X:000000H X:007FFFH XDATA X:000000H X:000000H X:007FFFH HDATA C:000000H C:000000H C:007FFFH 000438H CODE C:000000H C:000000H C:007FFFH CONST C:000000H C:000000H C:007FFFH ECODE B00:0000H C:000000H C:007FFFH HCONST I:000000H I:000000H I:0000FFH 000001H IDATA I:000000H I:000000H I:00007FH 00001CH DATA I:000020H.0 I:000020H.0 I:00002FH.7 000001H.1 BIT
And in particular:
000003H 00035EH 00035CH BYTE UNIT CODE ?PR?PRINTF?PRINTF
Which shows the original statement to be out by more than an order of magnitude.
Presuming the person who made this statement is reasonably familiar with C51, how are we supposed to have any confidence in the commentary on the unsuitability of C++ when it would appear that none of the contributors have actually used any of the available implementations? Should we assume that their 'experience' really is sufficient?
If I am wrong in my assumption that none of the contributors to this thread have used C++ on an 8051 please do correct me. If you can provide any actual data to support the hypothesis that C++ is a non-starter on an 8051 I would be genuinely interested.
Jack,
Finally someone responds with a positive point of view :)
Up to that time all responses have been leaning towards the "I know better than you because I've got experience" style.
What there seems to have been lacking is the desire to try something that they are unfamiliar with.
The words new, dog, tricks and old in a different order seem to be appropriate for some here.
Note that JS showed an example saying that the Keil C compiler + linker (the buggy tool you have been recommended to use) is quite good.
That should not be extrapolated into believing that the '51 is a processor well suited to C++.
There is still problems with dyanmic memory, virtual methods, templated code etc.
I did not say, nor have I assumed, that the 8051 is well suited to C++.
I am hoping (and increasingly believing) that it CAN be successfully used.
Some investigation I have done has revealed that there is even a Java VM that runs on a derivative (the Dallas 80C400).
Java is interpreted, C++ is compiled. Both have the problems that you mention.
So if (yet another) supplier provides tools for such a high level language to be used, it implies that there are requirements that can be satisfied with such tools.
"Some investigation I have done has revealed that there is even a Java VM that runs on a derivative (the Dallas 80C400)."
True.
"TINI is available online at TINI Store for $67.00 which includes 1 MByte SRAM and 512 KBytes of Flash ROM."
Tiny footprint...
At this point I can contribute some real-life experience.
I have developed a couple of applications using Java on the Dallas 80C400 - Their name for the technology is TINI.
I had previously done a lot of work (more than 15 years worth) primarily in assembler and C on 8051 and V55 cores.
TINI was my first attempt of using such a high level language on the 8051 derivative (albeit a vastly souped-up one).
The result - The projects were written in time, to cost and within the constraints laid down by the hardware.
Fortunately, they were not time critical applications and they performed their function adequately and (most importantly) within specification requirements.
Then came another project.
The management wanted me to use the same basic platform; i.e., the 80C400 with TINI. Knowing what the project entailed, I was reluctant to us TINI on this project but I had my orders and decided to go along with the decision.
It very quickly became apparent that the setup just was not man enough for the job. I decided (unknown to the management at that time) to rewrite the application in C.
The result was that the application ran some 400 time faster in various critical sections than the equivalent TINI code. Yes, I do actually mean 400 times faster!
Fortunately (for me), the management agreed that I had made the right decision.
So ... my advice is this:
Yes you CAN consider and use C, C++, JAVA or any other high level language, but please also consider whether the resultant application is going to work in a satisfactory and acceptable manner for the customer.