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.
hi: I want to use recursion function to transform a integer from binary form to ascii form. But the result is wrong.I don't why ,please help to analyse.
soure code:
void binary_to_ascii(unsigned int value) { unsigned int quotient; quotient = value/10; if(quotient !=0 ) binary_to_ascii(quotient); putchar(value%10+'0'); } int main() { while(1){ binary_to_ascii(4267); } }
I want to produce characters '4','2','6','7' in sequence. but the result is '4','4','4'. I don't know what is wrong with my code.
A recursive function requires a stack for storing the return values - check how large this stack must be for each iteration of a recursive function.
It is trivial to rewrite a very large group of recursive functions into iterative functions. In this case, the trivial implementation will produce the characters in reverse order - but each iteration only needs one byte to store one character until they are ready to be emitted in the correct order.
Recursion is a very useful concept. But it is best left for problems where the complexity of storing state information and making subdivision decisions gets very complex with a non-recursive method.
Any linear recursive algorithm can be trivially rewritten as an iteration. Many branching recursive algorithms (tree walking for example) can be trivially rewritten with an explicit state queue. It's quite seldom that a iterative solution can't be implemented in an obvious way.
It also requires some way to pass parameters & maintain local variables so that each "recursion" doesn't corrupt the data & parameters of previous recursions; ie, it requires that the functions are reentrant.
Most 'C' compilers achieve this by using the stack, so that their functions are inherently reentrant - Keil C51 does not do this and its functions are inherently not reentrant.
This is the same reason why Keil C51 has issues with function pointers - as previously explained: http://www.keil.com/forum/17469
Again, this is down to the specific features of the 8051 architecture and, again, you must not just assume that the 8051 is "just another processor" - it isn't!
"It also requires some way to pass parameters & maintain local variables so that each "recursion" doesn't corrupt the data & parameters of previous recursions; ie, it requires that the functions are reentrant."
Not always needed. The "standard" recursive function fac() who every school uses as prime example of recursion can work inplace. A lot of linear recursive functions just needs one set of registers.
to the OP
STOP what you are doing, read "the bible" get familiar with the architecture you are working with. You are trying to shoot down a 747 with a pellet gun.
Erik
here are the links to "the bible" Chapter 1 - 80C51 Family Architecture: www.nxp.com/.../80C51_FAM_ARCH_1.pdf
Chapter 2 - 80C51 Family Programmer’s Guide and Instruction Set: www.nxp.com/.../80C51_FAM_PROG_GUIDE_1.pdf
Chapter 3 - 80C51 Family Hardware Description: www.nxp.com/.../80C51_FAM_HARDWARE_1.pdf
A pellet in the eye of the pilot may be quite efficient - but having underpowered weapons means it takes much more planning and skill to manage.
I agree with the sage advice of the replys to this post but having said that, I must insist that recursion for the Keil C51 toolset does work well when you stay within it's limitations. You can set the max depth of recursion. And you limit the registers used by passing a minimum set of parameters. As posted, only registers R7 and R6 will be used to pass parameters. The pseudo stack is built "top down" in the Idata area. It's not difficult to see how much Idata area is used by the pseudo stack. Bradford
Keil C51 toolset does work well when you stay within it's limitations. and accept its achitecturally mandated inefficiencies.
Keil has, I am sure due to pressure fron the "C is C" types, implemented many functions that the architecture of the '51 makes extremely inefficient and likely to eat resources beyond what is available.
Sure, with a PC attitude, you can just implement all the inefficient code you want and then say "We need a faster peocessor"
Has anyone ever thought about the fact that while the greenies are constantly on the back of e.g. Dell, nobody complains when the next Microsoft wonder requires a much faster (and thus power hungry) computer.
But recursion is a valid programming paradigm.
And a 8051 can support recursion.
No reason to not use recursion for cases where it gives an advantage. Finding examples of how a tool can be abused doesn't mean that a tool should never be used.
no argument.
The point is that if you do "C is C" and do not know the implications with the architecture, you have not taken the time to get to know the '51 aka "architectural ignorance".
There are, indeed cases where a careful use of one of the features that have been wrested into C51 can be justified, but doing so just because "C is C" is plain foolish.
I kind of expect an embedded developer to be competent enough to: 1) Have a rough idea about mapping from C to assembler. 2) Be able to look at disassembly if expectations doesn't match reality.
It seems a very bad idea to put the "pseudo stack" into IDATA!
IDATA is only 128 or 256 bytes - and also has to accomodate the "real" hardware stack...
This case is not one of them - especially on an 8051!
Sorry - that post should have been titled, "use recursion for cases where it gives an advantage"
In other words: I agree that recursion may, in general, be used where it gives an advantage (within the limitations of the implementation) - but the example at hand is not one of them!
http://www.keil.com/support/man/docs/c51/c51_le_reentrantfuncs.htm
* Small model reentrant functions simulate the reentrant stack in idata memory. * Compact model reentrant functions simulate the reentrant stack in pdata memory. * Large model reentrant functions simulate the reentrant stack in xdata memory.
Yes, I know it can be done; what I said was that I thought it was a bad idea to do it - for the reasons stated!
You are right. But in general I think, an unlimited recursion is a bad idea. In other case, the maximum size of the required stack is known.