www.digitalmars.com         C & C++   DMDScript  

digitalmars.D.learn - There must be a better way

reply "Emp" <empty nomail.com> writes:
I wrap two values (x && y coordinates) like this:

uint wrap(uint axis, int value)
{
 int max=0;
 if (axis==1) max=25;
 if (axis==0) max=10;
 if(value>=max){
  return (value % max);
 }
 if(value<0){
  int newValue;
  newValue=value;
  while(newValue<0){
   newValue+=max;
  }
  return (newValue);
 }
return value;
}

So I need to do things like 'wrap(0,currentX)' everytime to wrap the
currentX. Like:

(the original x && y need to be untouched)
array[wrap(0,currentX)][wrap(1,currentY)];

Is this really the best way, or am I just missing some nifty D
programming?

Two small questions:
1. Isn't inout more used than out for functions, in contrast to what the
website says?
    I use inout quite often, or am I just doing something wrong?

2. Is it really difficult to make '>=" correctly compare an unsigned and a
signed int?
    (This took me some time to find out :)

Grtz, Emp
Aug 01 2006
parent reply Derek Parnell <derek nomail.afraid.org> writes:
On Wed, 2 Aug 2006 03:57:07 +0200, Emp wrote:

 I wrap two values (x && y coordinates) like this:
 
 uint wrap(uint axis, int value)
 {
  int max=0;
  if (axis==1) max=25;
  if (axis==0) max=10;
  if(value>=max){
   return (value % max);
  }
  if(value<0){
   int newValue;
   newValue=value;
   while(newValue<0){
    newValue+=max;
   }
   return (newValue);
  }
 return value;
 }
 
 So I need to do things like 'wrap(0,currentX)' everytime to wrap the
 currentX. Like:
 
 (the original x && y need to be untouched)
 array[wrap(0,currentX)][wrap(1,currentY)];
 
 Is this really the best way, or am I just missing some nifty D
 programming?

Here is an alternative... const AXIS0 = 10; const AXIS1 = 25; int mwrap(int max, int value) { if(value >= max) { value %= max; } else if (value < 0) { value = max + (value % max); if (value == max) value = 0; } return value; } unittest { assert(mwrap(AXIS0, 35) == 5); assert(mwrap(AXIS0, 0) == 0); assert(mwrap(AXIS0, 6) == 6); assert(mwrap(AXIS0, 10) == 0); assert(mwrap(AXIS1, 35) == 10); assert(mwrap(AXIS1, 0) == 0); assert(mwrap(AXIS1, 6) == 6); assert(mwrap(AXIS1, 25) == 0); assert(mwrap(AXIS0, -35) == 5); assert(mwrap(AXIS0, -6) == 4); assert(mwrap(AXIS0, -10) == 0); assert(mwrap(AXIS1, -35) == 15); assert(mwrap(AXIS1, -6) == 19); assert(mwrap(AXIS1, -25) == 0); } The main change I made was to supply the actual maximum value as a parameter. This will make the function more flexible in future. The other change was to not a D trick, just a maths 'trick'.
 Two small questions:
 1. Isn't inout more used than out for functions, in contrast to what the
 website says?
     I use inout quite often, or am I just doing something wrong?

Yes, IMHO <g> The use of return values rather than updating the input parameters leads to programs that are easier to maintain and re-use.
 2. Is it really difficult to make '>=" correctly compare an unsigned and a
 signed int?

No. This is huge wart in D. For some Bob-only-knows reason, D silently interprets the bit-value of an int as if was a uint when doing such comparisons. Daft! If you know that there is not going to be an overflow issue, you can do this ... if (cast(int)my_uint >= my_int) ... which is saying you want the uint converted to a signed value before comparing the two. -- Derek (skype: derek.j.parnell) Melbourne, Australia "Down with mediocrity!" 2/08/2006 12:41:01 PM
Aug 01 2006
parent reply "Emp" <empty nomail.com> writes:
Thanks for the maths :)
The maximum are not constant and writing out the place they live would yield 
to something like:

array[wrap(something[var].maxX,currentX)][wrap(something[var].maxY,currentY)];

So the int was a bit of a hack... sorry :)

About the inout:
How would you do something like this?

bool something(inout structure struc ,int var){
  for (int i=0; i < struc.data[].length; i++){
    if(struc.data[i].count==0){
      struc.data[i].type=var;
      struc.data[i].count=30;
      return true;
    }
  }
  return false;
}
Aug 01 2006
next sibling parent Derek Parnell <derek nomail.afraid.org> writes:
On Wed, 2 Aug 2006 05:49:44 +0200, Emp wrote:

 About the inout:
 How would you do something like this?
 
 bool something(inout structure struc ,int var){
   for (int i=0; i < struc.data[].length; i++){
     if(struc.data[i].count==0){
       struc.data[i].type=var;
       struc.data[i].count=30;
       return true;
     }
   }
   return false;
 }

Ah yes... the problem with structs. The way you have done it here is a trade-off for performance and works fine. The alternative would be to pass back and forth the complete structure which is not a generally a good idea. You could make it a class instead of a structure but that would be just pedantic ;-) -- Derek (skype: derek.j.parnell) Melbourne, Australia "Down with mediocrity!" 2/08/2006 2:39:58 PM
Aug 01 2006
prev sibling parent reply "Unknown W. Brackets" <unknown simplemachines.org> writes:
I'm not clear on where your going, but I like to keep things simpler and 
thus more maintainable.  Instead of:

array[wrap(something[var].maxX,currentX)][wrap(something[var].maxY,currentY)];

I would probably prefer...

whatever_t getWrapped(whatever_t[][] array, int x, int y, int var)
{
	return array[wrap(something[var].maxX, x)][wrap(something[var].maxY, y)];
}

Then you'd do:

array.getWrapped(currentX, currentY, var);

Which would seem much easier, and should be optimized out the same with 
inlining.  But this might not be practical depending on what "something" 
is (I'm guessing here it's a lookup or something.)

Also, fwiw, I use inout all the time.  I think there are specific design 
patterns and code paths with which it makes complete sense.  Example:

// Attempt to bring item to the top/head of the linked list.
if (!bringToTop(linked_list, item))
	writefln("Uh oh, %s was not found!", item.toString());

I don't think it's ambiguous that linked_list might be modified.  Just 
my opinion.  I might prefer "linked_list.bringToTop(item)" if it made 
sense, though (since that's even harder to misunderstand.)

-[Unknown]


 Thanks for the maths :)
 The maximum are not constant and writing out the place they live would yield 
 to something like:
 
 array[wrap(something[var].maxX,currentX)][wrap(something[var].maxY,currentY)];
 
 So the int was a bit of a hack... sorry :)
 
 About the inout:
 How would you do something like this?
 
 bool something(inout structure struc ,int var){
   for (int i=0; i < struc.data[].length; i++){
     if(struc.data[i].count==0){
       struc.data[i].type=var;
       struc.data[i].count=30;
       return true;
     }
   }
   return false;
 }
 
 
 
 
 
 
 

Aug 02 2006
parent reply "Emp" <empty nomail.com> writes:
I've been looking through your post but kind quite grasp how I should prog 
like that... :/
I want to do stuff like this:
(I hope it is a bit more clear)

type=block.data[wrap(0,x)][wrap(1,y)].type;

with:
uint wrap(uint axis, int value)
{
 int max=1;
 if (axis==1) max=block.maxX; //variable
 if (axis==0) max=block.maxY; //same :)

 if(value >= max){
        value %= max;
    }else if (value < 0){
  value = max + (value % max);
  if (value == max) value = 0;
    }

return value;
}

 I'm not clear on where your going, but I like to keep things simpler and 
 thus more maintainable.  Instead of:

 array[wrap(something[var].maxX,currentX)][wrap(something[var].maxY,currentY)];

 I would probably prefer...

 whatever_t getWrapped(whatever_t[][] array, int x, int y, int var)
 {
 return array[wrap(something[var].maxX, x)][wrap(something[var].maxY, y)];
 }

 Then you'd do:

 array.getWrapped(currentX, currentY, var);

 Which would seem much easier, and should be optimized out the same with 
 inlining.  But this might not be practical depending on what "something" 
 is (I'm guessing here it's a lookup or something.)

 Also, fwiw, I use inout all the time.  I think there are specific design 
 patterns and code paths with which it makes complete sense.  Example:

 // Attempt to bring item to the top/head of the linked list.
 if (!bringToTop(linked_list, item))
 writefln("Uh oh, %s was not found!", item.toString());

 I don't think it's ambiguous that linked_list might be modified.  Just my 
 opinion.  I might prefer "linked_list.bringToTop(item)" if it made sense, 
 though (since that's even harder to misunderstand.)

 -[Unknown]


 Thanks for the maths :)
 The maximum are not constant and writing out the place they live would 
 yield to something like:

 array[wrap(something[var].maxX,currentX)][wrap(something[var].maxY,currentY)];

 So the int was a bit of a hack... sorry :)

 About the inout:
 How would you do something like this?

 bool something(inout structure struc ,int var){
   for (int i=0; i < struc.data[].length; i++){
     if(struc.data[i].count==0){
       struc.data[i].type=var;
       struc.data[i].count=30;
       return true;
     }
   }
   return false;
 }





 


Aug 03 2006
parent "Unknown W. Brackets" <unknown simplemachines.org> writes:
I guess, then, you'd have block with data inside it, as you do now, but 
data would have to be a class.  It would have to know about its parent, 
block.

Then you would be able to do something like that.

However, if you can settle for:

type = block.data(wrap(0, x), wrap(1, y)).type;

Then you don't need a class to proxy things.  But, is the call to wrap 
mandatory?  It looks like you really want something like below...

Let me note that D, when it sees this:

x.y(z);

Will try:

y(x, z);

If the function is available.  That's what I'm using (abusing?) here.

data_t data(whatever_block_is_t block, uint x, uint y)
{
	return return block.realData[wrap(x, block.maxX)][wrap(y, block.maxY)];
}

uint wrap(uint value, uint max)
{
	if (value >= max)
		return value % max;
	else if (value < 0)
		return max + (value % max);
	else
		return value;
}

So then you'd do:

type = block.data(x, y).type;

Assuming you always want the x and why wrapped.  If you didn't, you 
could still do:

type = block.realData[x][y].type;

-[Unknown]


 I've been looking through your post but kind quite grasp how I should prog 
 like that... :/
 I want to do stuff like this:
 (I hope it is a bit more clear)
 
 type=block.data[wrap(0,x)][wrap(1,y)].type;
 
 with:
 uint wrap(uint axis, int value)
 {
  int max=1;
  if (axis==1) max=block.maxX; //variable
  if (axis==0) max=block.maxY; //same :)
 
  if(value >= max){
         value %= max;
     }else if (value < 0){
   value = max + (value % max);
   if (value == max) value = 0;
     }
 
 return value;
 }
 
 I'm not clear on where your going, but I like to keep things simpler and 
 thus more maintainable.  Instead of:

 array[wrap(something[var].maxX,currentX)][wrap(something[var].maxY,currentY)];

 I would probably prefer...

 whatever_t getWrapped(whatever_t[][] array, int x, int y, int var)
 {
 return array[wrap(something[var].maxX, x)][wrap(something[var].maxY, y)];
 }

 Then you'd do:

 array.getWrapped(currentX, currentY, var);

 Which would seem much easier, and should be optimized out the same with 
 inlining.  But this might not be practical depending on what "something" 
 is (I'm guessing here it's a lookup or something.)

 Also, fwiw, I use inout all the time.  I think there are specific design 
 patterns and code paths with which it makes complete sense.  Example:

 // Attempt to bring item to the top/head of the linked list.
 if (!bringToTop(linked_list, item))
 writefln("Uh oh, %s was not found!", item.toString());

 I don't think it's ambiguous that linked_list might be modified.  Just my 
 opinion.  I might prefer "linked_list.bringToTop(item)" if it made sense, 
 though (since that's even harder to misunderstand.)

 -[Unknown]


 Thanks for the maths :)
 The maximum are not constant and writing out the place they live would 
 yield to something like:

 array[wrap(something[var].maxX,currentX)][wrap(something[var].maxY,currentY)];

 So the int was a bit of a hack... sorry :)

 About the inout:
 How would you do something like this?

 bool something(inout structure struc ,int var){
   for (int i=0; i < struc.data[].length; i++){
     if(struc.data[i].count==0){
       struc.data[i].type=var;
       struc.data[i].count=30;
       return true;
     }
   }
   return false;
 }



Aug 03 2006