Writing a incrementing/decrementing function with roll-over

[pozz]

Hi all,

I need to write a simple incrementing/decrementing function like this:

unsigned char
change( unsigned char x, unsigned char min, unsigned char max, signed char d);

x is the value to increase/decrease
min is the minimum value that x can assume
max is the maximum value that x can assume
d is positive or negative and indicates how much x must be incremented or
decremented
Of course, the return value is the new value for x.

Some examples to clarify the question:
x=10,min=0,max=20,d=5 --> 15
x=10,min=0,max=20,d=-5 --> 5
x=18,min=0,max=20,d=5 --> 2
x=18,min=5,max=20,d=5 --> 7
x= 3,min=0,max=20,d=-5 --> 19
x=10,min=8,max=20,d=-5 --> 18

I'd like a function that use only unsigned char...



Another similar question. I have a variabile x (unsigned char) and I must
increment it. The increment is stored in another variable, d (unsigned char).
x can't assume values greater than max, stored in another variable (unsigned
char).
I usually write:

if( x+d>max )
x = max;
else
x += d;

I think it's not correct because what happens when x=200, d=100 and max=220??
If I write:

if( x>max-d )
x = max;
else
x += d;

what happens when max=10 and d=20?? I need a variable greater than unsigned
char to do the temporary sum x+d? And if I use long, I need extra-long
variables?


Thank you very much for the help.

 

[Dann Corbit]

If you use wider types (with unsigned char to be the final target), it will
be easier.

E.g. use unsigned int instead of unsigned char.

 

[Fred Kleinschmidt]

Hi all,

I need to write a simple incrementing/decrementing function like this:

unsigned char
change( unsigned char x, unsigned char min, unsigned char max, signed char
d);

x is the value to increase/decrease
min is the minimum value that x can assume
max is the maximum value that x can assume
d is positive or negative and indicates how much x must be incremented
or
decremented
Of course, the return value is the new value for x.

Some examples to clarify the question:
x=10,min=0,max=20,d=5 --> 15
x=10,min=0,max=20,d=-5 --> 5
x=18,min=0,max=20,d=5 --> 2
x=18,min=5,max=20,d=5 --> 7
x= 3,min=0,max=20,d=-5 --> 19
x=10,min=8,max=20,d=-5 --> 18

I'd like a function that use only unsigned char...



Another similar question. I have a variabile x (unsigned char) and I must
increment it. The increment is stored in another variable, d (unsigned
char).
x can't assume values greater than max, stored in another variable
(unsigned
char).
I usually write:

if( x+d>max )
x = max;
else
x += d;

I think it's not correct because what happens when x=200, d=100 and
max=220??
If I write:

if( x>max-d )
x = max;
else
x += d;

what happens when max=10 and d=20?? I need a variable greater than
unsigned
char to do the temporary sum x+d? And if I use long, I need extra-long
variables?

x = ( (d > max) || (x > max-d) ) ? max : x+d;

 

[goose]

Hi all,

I need to write a simple incrementing/decrementing function like this:

unsigned char
change( unsigned char x, unsigned char min, unsigned char max, signed char d);

x is the value to increase/decrease
min is the minimum value that x can assume
max is the maximum value that x can assume
d is positive or negative and indicates how much x must be incremented or
decremented
Of course, the return value is the new value for x.

Some examples to clarify the question:
x=10,min=0,max=20,d=5 --> 15
x=10,min=0,max=20,d=-5 --> 5
x=18,min=0,max=20,d=5 --> 2
x=18,min=5,max=20,d=5 --> 7
x= 3,min=0,max=20,d=-5 --> 19
x=10,min=8,max=20,d=-5 --> 18

I'd like a function that use only unsigned char...

I won't directly help with this, but see below for a hint or two.

Another similar question. I have a variabile x (unsigned char) and I must
increment it. The increment is stored in another variable, d (unsigned char).
x can't assume values greater than max, stored in another variable (unsigned
char).
I usually write:

You can normally do this:
x = (x + d) % (max+1);

Now using that, you should be able to figure
out a way to solve your problem above, right?

Post your attempt if it won't work

<snipped>

 

[spibou]

Hi all,

I need to write a simple incrementing/decrementing function like this:

unsigned char
change( unsigned char x, unsigned char min, unsigned char max, signed char d);

x is the value to increase/decrease
min is the minimum value that x can assume
max is the maximum value that x can assume
d is positive or negative and indicates how much x must be incremented or
decremented
Of course, the return value is the new value for x.

Some examples to clarify the question:
x=10,min=0,max=20,d=5 --> 15
x=10,min=0,max=20,d=-5 --> 5
x=18,min=0,max=20,d=5 --> 2
x=18,min=5,max=20,d=5 --> 7
x= 3,min=0,max=20,d=-5 --> 19
x=10,min=8,max=20,d=-5 --> 18

I'd like a function that use only unsigned char...

Hint: Solve the problem first when min=0. Make use
of the % operator.

Another similar question. I have a variabile x (unsigned char) and I must
increment it. The increment is stored in another variable, d (unsigned char).
x can't assume values greater than max, stored in another variable (unsigned
char).
I usually write:

if( x+d>max )
x = max;
else
x += d;

I think it's not correct because what happens when x=200, d=100 and max=220??
If I write:

Are you worried that x+d may be larger than UCHAR_MAX ?
If that's the problem then there is an easy solution:

if ( x + d < x || x + d > max ) x=max ;
else x += d ;

The point is that x+d has overflowed if and only if
x+d < x

Two remarks about the solution above:
1) It assumes that min=0. It will be easy to modify
it for when min>0.
2) It works for all kinds of unsigned integer types. The
only assumption is that x,d,max are all expressed in the
same unsigned integer type.

what happens when max=10 and d=20?? I need a variable greater than unsigned
char to do the temporary sum x+d? And if I use long, I need extra-long
variables?

No , you don't need to use larger "variables" than what x and d
are. You could solve the problem by using larger types but
what if the problem specified that you have to work with
unsigned long long ? You'd be in trouble then.

If you use wider types (with unsigned char to be the final target), it will
be easier.

E.g. use unsigned int instead of unsigned char.

Is unsigned int guaranteed to be wider than unsigned char ?

You can normally do this:
x = (x + d) % (max+1);

This assumes that min=0 , right ?


Spiros Bousbouras

 

[pozz]

goose ha scritto:

You can normally do this:
x = (x + d) % (max+1);

This is my attempt for the change function with roll-over:

unsigned char
change_roll( unsigned char x, unsigned char min, unsigned char max,
signed char d )
{
if( d>0 ) {
x -= min;
max -= min;
return (x+d)%(max+1) + min;
} else {
/* ??? */
}
}

What do I write in the else branch when d is negative?

And this is my attempt for the change function without roll-over:

unsigned char
inc( unsigned char x, unsigned char min, unsigned char max, signed char
d )
{
if( d>0 && (x+d<x || x+d>max) )
return max;
else if( d<0 && (x+d>x || x+d<min) )
return min;
else
return x+d;
}


What do you say about that?

 

[spibou]

And this is my attempt for the change function without roll-over:

unsigned char
inc( unsigned char x, unsigned char min, unsigned char max, signed char
d )
{
if( d>0 && (x+d<x || x+d>max) )
return max;
else if( d<0 && (x+d>x || x+d<min) )
return min;
else
return x+d;
}


What do you say about that?

You said in your first post that for the "without roll-over" part d is
an *unsigned* char but your definition says that d is a signed char.
Still , it looks correct to me.

Spiros Bousbouras

 

[pozz]

(e-mail address removed) ha scritto:

You said in your first post that for the "without roll-over" part d is
an *unsigned* char but your definition says that d is a signed char.

Oh, you are right. I generalized the function for increment and
decrement.

Still , it looks correct to me.

Hmm..., I rethought about that and I worry for the following line:
return (x+d)%(max+1) + min;
What happens if x=200, d=200, max=210, min=0 and the architecture uses
only byte (8-bit
microcontroller application)? The return value will be 144 but it is
wrong!

 

[spibou]

goose ha scritto:


This is my attempt for the change function with roll-over:

unsigned char
change_roll( unsigned char x, unsigned char min, unsigned char max,
signed char d )
{
if( d>0 ) {
x -= min;
max -= min;
return (x+d)%(max+1) + min;
} else {
/* ??? */
}
}

What do I write in the else branch when d is negative?

It's not correct I'm afraid. Consider the case where
min=0 , max=199 , UCHAR_MAX=255 , x=199 , d=57

Then your function ought to return 56 but it will return 0.

Spiros Bousbouras

 

[spibou]

(e-mail address removed) ha scritto:


Oh, you are right. I generalized the function for increment and
decrement.



Hmm..., I rethought about that and I worry for the following line:
return (x+d)%(max+1) + min;
What happens if x=200, d=200, max=210, min=0 and the architecture uses
only byte (8-bit
microcontroller application)? The return value will be 144 but it is
wrong!

When I said "it looks correct to me" I was referring *only* to the code
I
quoted. I still believe it is correct. But as you noticed yourself the
other
piece of code is not correct although your counterexample is not
strictly
correct either. That's because you have declared d as signed char so if
you only have 8 bits then it cannot have the value 200. Nevertheless I
provided a counterexample in a different post.

You have to think about some elementary number theory with regards
to remainders in order to arrive at a correct function.

Spiros Bousbouras

 

[pozz]

(e-mail address removed) ha scritto:

It's not correct I'm afraid. Consider the case where
min=0 , max=199 , UCHAR_MAX=255 , x=199 , d=57

Then your function ought to return 56 but it will return 0.

Yes, you are right. I must check if (x+d) generates an overflows
(x+d<x).

if( d>=0 ) {
x -= min;
max -= min;
if( x+d<x )
return d - (max-x) + min;
else
return (x+d)%(max+1) + min;
} else
????

But what should I write when d is negative?