Thread: stuck : conways game of life (lots of code)

Originally Posted by anduril462

I'm with Quzah here. Unless you can show me the specs from your assignment that say empty rows have to be null, then they don't. And they shouldn't be. It just complicates things. You can have a column with all 0's down it if there are no creatures living anywhere in that column. Why can't you have a row with all 0 elements if nothing lives there?

Fine :


And I've been posting the code which has a description of what the functions should do on this forum since the beginning which all clearly state:

Originally Posted by anduril462

So compile that for me, with debugging symbols (gcc -g life.c). Then, run it in gdb and find the exact line where it fails. If you need a gdb tutorial, check here: Cprogramming.com - Tutorials - An Introduction to GDB

Actually, we're really close with the rotate code. We somehow managed to rotate -90 degrees (1/4 turn counter-clockwise) instead.

Sorry, but I don't think it will be particularly useful for me to compile that as it needs to work with nulls as I keep saying. I know how to use GDB, I've been using it for this thing for ages, its how I know about nulls via print(blah) etc. and back trace (bt).. etc. though it doesn't take much to see where it fails.. every time the same line. new[blah]=old[blah]

Thanks for doing the above though it is interesting looking at how you're trying to rotate it.

You have an entirely different problem then.

When your evolve() and rotate() functions return, any unoccupied columns should be represented by NULLs in the arena array rather than by arrays of characters.

Actually, you have TWO new problems with this horrible design.

1. If I have this:

A C

E G

When I rotate that 90 degrees, I now have a new column that is null.

E A

G C


2. He's telling you to null out COLUMNS, not rows. That means he wants column-major, not row-major arrays. That's completely ass backwards.

Anyway, fix both of these problems at the same time, by doing what I suggested first, except flip X and Y since your professor is an idiot, and pretend rows are columns instead of rows.

Build yourself a function that allocates rows (columns) first, and then each length of columns (rows). Now copy all of the contents of the old one into its place. Now run through each row (column) and if there are no values in it, free that row (column) and set that pointer to NULL.


90 degree clockwise rotation should be easy then: new[ MAXY - 1 - Y ][ X ] = old[ X ][ Y ]
I think that's right. I had it figured out until your professor didn't know what a row and a column was anymore.


Quzah.

Originally Posted by quzah

2. He's telling you to null out COLUMNS, not rows. That means he wants column-major, not row-major arrays. That's completely ass backwards.

yes, I tell me about it...lol you can see why I'm getting so confused with it all

Originally Posted by quzah

Anyway, fix both of these problems at the same time, by doing what I suggested first, except flip X and Y since your professor is an idiot, and pretend rows are columns instead of rows.

Is this what you mean by this:

Code:

	for(y = 0; y < *pheight; y++)
	{
		for(x =0; x < *pwidth; x++)
		{
			...
		}
		
	}

Or is that what you meant by the bit below? Or did you mean by the below you were talkiing about changing:

Code:

newarena = malloc((*pheight) * sizeof(char*));
	for(x = 0; x < *pwidth; x++)
	{
			newarena[x] = malloc((*pwidth) * sizeof(*newarena[x]));
	}

Originally Posted by quzah

Build yourself a function that allocates rows (columns) first, and then each length of columns (rows). Now copy all of the contents of the old one into its place.

I understand this bit: Now run through each row (column) and if there are no values in it, free that row (column) and set that pointer to NULL. <<Actually, should we really free(arena[y]) before we have returned the newarena or are you suggesting I do that after the function has been called in the main function?

I think your wording may be confusing me on this. Build a function... allocates rows first and then length of columns...

Originally Posted by quzah

90 degree clockwise rotation should be easy then: new[ MAXY - 1 - Y ][ X ] = old[ X ][ Y ]
I think that's right. I had it figured out until your professor didn't know what a row and a column was anymore.
this also makes sense to me, although I'm not entirely sure whether it will work.. will try
Quzah.

Sorry the wording is confusing me a little and my brain is a bit fried and worried whether I'll manage this project or not.

Make a function that allocates a new grid:

Code:

char**newgrid( int cols, int rows )
{
    char **g = NULL;
    int c = 0;
    g = malloc( y * sizeof *g );
    for( c = 0; c < cols; c++ )
        g[ c ] = malloc( r * sizeof **g );
    return g;
}

I did no error checking there, normally you would. Anyway, now that you can make grids that you want of any size, after you've copied the new contents over top of it:

Code:

void foo( char **g, int c, int r )
{
    for each column
        for each row
            if( g[ c ][ r ] == notempty ) break;
        if r == max rows
            we made it to the end of a column without encountering anything valid
            free( g[ c ] )
            g[ c ] = NULL
}

Quzah.
}

Originally Posted by quzah

Make a function that allocates a new grid:

Code:

char**newgrid( int cols, int rows )
{
    char **g = NULL;
    int c = 0;
    g = malloc( y * sizeof *g );
    for( c = 0; c < cols; c++ )
        g[ c ] = malloc( r * sizeof **g );
    return g;
}

I did no error checking there, normally you would. Anyway, now that you can make grids that you want of any size, after you've copied the new contents over top of it:

Code:

void foo( char **g, int c, int r )
{
    for each column
        for each row
            if( g[ c ][ r ] == notempty ) break;
        if r == max rows
            we made it to the end of a column without encountering anything valid
            free( g[ c ] )
            g[ c ] = NULL
}

Quzah.
}

Is there a way we could do this in one function as I'm not sure we can make our own functions... I can see how todo the first bit maybe

Code:

char** newarena = NULL;
	int i = 0;
    	newarena = malloc( *pwidth * sizeof(*newarena));
    	for( i = 0; i < cols; c++ )
        newarena[ i ] = malloc( *pheight * sizeof(**newarena));

perhaps

Sure. Just attach the loops in the bottom function to the portion of that function after you have copied the contents over.


Quzah.

Originally Posted by quzah

Sure. Just attach the loops in the bottom function to the portion of that function after you have copied the contents over.


Quzah.

Oh okay well I didn't try to do it because I couldn't get my head around how to deal with the break bit etc.

Is this OK in theory?

Code:

code removed..

You can't just free(arena) there. You have to free each column (since you're doing it backwards) first. You need one free call for every associated *alloc call you have made.


Quzah.

Originally Posted by play_fetch

The freeing arena is the old arena passed to the rotate method so surely that is okay as it isn't needed any more? I can't free anything (relating to newarena) until I return it anyway can I?

You aren't understanding:

Code:

char**r;
r = malloc( 3 * sizeof *r ); /*1*/
for( x = 0; x < 3; x++ )
    r[x] = malloc( 4 * sizeof **r ); / 3 more */

You need four free calls to properly free that:

Code:

for( x = 0; x < 3; x++ )
    free( r[x] );
free( r );

You can't just free( r ), you have to free each column r contains first.


Quzah.

Originally Posted by quzah

You aren't understanding:

Code:

char**r;
r = malloc( 3 * sizeof *r ); /*1*/
for( x = 0; x < 3; x++ )
    r[x] = malloc( 4 * sizeof **r ); / 3 more */

You need four free calls to properly free that:

Code:

for( x = 0; x < 3; x++ )
    free( r[x] );
free( r );

You can't just free( r ), you have to free each column r contains first.


Quzah.

So you're saying I need to loop through each x < *pwidth for arena[x] and free it? In the main code and then free(arena)?

I'm sorry I don't catch on so quick but I get confused easily with all the rows and cols being different etc.

Code:

code removed

That's why I was advocating separate functions to allocate and free your grids. You could then just do:

Code:

newgrid = getgrid( newcols, newrows );
... do rotate ...
freegrid( oldgrid, oldsize );

For one, your code looks cleaner. Plus you know that once you have the function that works correctly, you can just use it without having to recheck all your new code, and make sure you didn't forget anything.


Quzah.

Well I'm going to be doing similar things for evolve I imagine, so a separate function is good to have around.. I may need to move it to another c file. They said it was OK to create another .c file, so, I assume functions are OK too.

So this code below looks OK (except I know it isn't perfect due to lots of error messages)? I'm getting errors about implicit function calls and such.. not sure what it means by that I'm not familiar with that error.