[karajorma]

Does that mean I've got permission to start adding the stuff from the patch in the Diaspora internal to trunk then?

[WMCoolmon]

Check your PMs.

[Flipside]

This was a rather interesting approach to checking four points surrounding a position on a grid as part of an A* Pathfinding setup...

Code: [Select]

        for (yPos = -1; yPos >1; yPos = yPos + 2)
        {
                    for (xPos = -1; xPos >1; xPos = xPos = xPos+2)
                    {
                        xPosB = (xPos + (path[pointer].getXValue()));
                        yPosB = (yPos + (path[pointer].getXValue()));
                        current = new XYValue(xPosB, 0);
                        if (roads.getByLocation(current) != 0)
                        {
                            options.add(current);
                        }
                        current = new XYValue(0,yPosB);
                        if (roads.getByLocation(current) != 0)
                        {
                            options.add(current);
                        }
                    }
        }
From that point onwards, I decided to do it the old-fashioned way

[blackhole]

Never, ever, ever convert an unsigned long to a double and then back again if you have 4294967295 assigned to it.

[lx]

a few years ago i tried to write a class which simulates an integer, except that this integer could handle much larger values than a normal 4-byte integer.
naturally i had to implement all the basic arithmetic functions (+,-,*,/)....
a few months ago i found my code again and wanted to see how i solved this stuff... and i could figure out everything except my division function (which works correct and surprisingly(!) fast)... but i have no f*king clue how it works and what i did there *g*
also it seems that at that time i thought it's obvious how it works and just refused to comment it .... d'oh.

maybe it was a result of the Ballmer Peak.. i don't know

so, here it is, that ugly beast ... :

Code: [Select]

BigInt& BigInt::operator/=(BigInt divisor)
{
const BigInt biNull("0");
if (divisor==biNull)
throw exception("division through zero");
   
if (/*this*/isLess/*than*/(divisor)) {
// result would be 0.xxxxx
num_.clear();
num_.push_back('0');
return *this;
}

cVec result;
vsize div_len = divisor.length();

while (divisor<=*this)
divisor << 1; // fast divisor*10
divisor >> 1;

unsigned i=1, k=0, l=0;
while (true) {
while (divisor>*this) {
if (*this==biNull || (i>=2 && l==k)) {
result.insert(result.begin(), '0');
i=1;
}
divisor >> 1; // fast divisor/10
++i;
l=k;
}
++k;
if (divisor.length() < div_len)
break;

unsigned j=0;
for (; divisor<=*this; ++j)
*this -= divisor;

result.insert(result.begin(), j+'0');
}
if (result.size()>1 && *result.begin() == '0')
result.erase(result.begin());

num_.swap(result);

return *this;
}
maybe i will give it a second look now.. with a pen and a paper

[lx]

Browsing through the source files of my current project I found this brilliant piece code.

Converting a number to a string, the easy way:

Code: [Select]

string ObjectManager::CreateStringFromNumber(unsigned int value, unsigned int digit)
{
   #if (!((('9' - 9) == '0') && (('8' - 8)8) == '0') && (('7' - 7) == '0') && (('6' - 6) == '0') && (('5' - 5) == '0') && (('4' - 4) == '0') && (('3' - 3) == '0') && (('2' - 2) == '0') && (('1' - 1) == '0')))
   Logger log("CreateStringFromNumber");
   log.WriteLine(RootLogger::ERROR, "Failed to convert from int to char");
   throw "Error";
   #endif


   string result(digits, '0');
   int tmp = 1;

   for (unsigned int i = 1; i < digits; ++i)
   {
   tmp *= 10;
   }

   for (string::iterator iter = result.begin(); iter != result.end(); ++iter)
   {
   (*iter) = static_cast<char>((value/tmp)+'0');
   value %= tmp;
   tmp /= 10;
   }
   return result;
}
i guess sprintf just doesn't cut it any more, right uchuujin

[Uchuujinsan]

get lost :>

[blackhole]

Code: [Select]

void cUsable::Use();
Now my code really is on drugs.

[portej05]

Thought I'd post one from within the SCP itself.
/analyse gives the rather tame

Code: [Select]

3>c:\users\james\documents\visual studio 2008\projects\freespace2\code\graphics\generic.cpp(36) : warning C6054: String 'argument 1' might not be zero-terminated: Lines: 32, 35, 36

The code that caused this warning?

Code: [Select]

ga->filename[strlen(ga->filename)] = '\0';


[Aardwolf]

portej05:

That line makes no sense... strlen(ga->filename) won't work right if it isn't null-terminated, but assuming it doesn't crash from an access violation, the return value will have '\0' there already...

 

[portej05]

Aardwolf: I know Hence why it is here

[taylor]

That is just beyond wrong.  And it has gotten much worse after seeing it since it appears that was part of one of my commits.

[portej05]

Just found a pile of these:
orienteditor.cpp Line 220:

Code: [Select]

m_spin1.SetRange((short)99999, (short)-99999);

For those who aren't sure what's going on, the maximum value of a signed short is 32767

[voidSkipper]

Some exercises in unintentional obfuscation from my game dev library.

Code: [Select]

if (o.x < cx - allow) {
safePush(c - 1, r, o);
}
if (o.x > cx+allow) {
safePush(c + 1, r, o);
}
if (o.y < cy - allow) {
if (o.x < cx - allow) {
safePush(c - 1, r - 1, o);
}
if (o.x > cx+allow) {
safePush(c + 1, r - 1, o);
}
safePush(c, r - 1, o);
}
if (o.y > cy + allow) {
if (o.x < cx - allow) {
safePush(c - 1, r + 1, o);
}
if (o.x > cx+allow) {
safePush(c + 1, r + 1, o);
}
safePush(c, r + 1, o);
}

safePush(c, r, o);

Code: [Select]

for (var c:int = 0; c < density; c++) {
for (var r:int = 0; r < density; r++) {
for (i = 0; i < grid[c][r].length; i++) {
var a:Entity = grid[c][r][i];
for (var j:int = i + 1; j < grid[c][r].length; j++) {
if (!a.checkDupeColl(grid[c][r][j]) && !grid[c][r][j].checkDupeColl(a)) {
a.collided.push(grid[c][r][j]);
grid[c][r][j].collided.push(a);
if (a.collisions.willHit(grid[c][r][j].collisions.type)) {
a.entityCollisionSolve(grid[c][r][j], true);
}
}
}
}
}
}

[Mongoose]

That is some fantastic use of whitespace and descriptive variable names.

[voidSkipper]

It made sense at the time.

Actually, and sadly, it still makes sense now.

The whitespace is because flashdevelop indents in tabs, and this was already tabbed in a great deal due to class/package/other functions.

[Sushi]

It made sense at the time.

Actually, and sadly, it still makes sense now.

The whitespace is because flashdevelop indents in tabs, and this was already tabbed in a great deal due to class/package/other functions.

The problem isn't the tabs: the problem is having a tab spacing of 8. Cut that to 4 (or 2, if you prefer) and you'll be a much happier chipmunk.

[voidSkipper]

But... I like it like this ):

[castor]

Code: [Select]

if (o.x < cx - allow) {
safePush(c - 1, r, o);
}
if (o.x > cx+allow) {
safePush(c + 1, r, o);
}
Doh! quite generous spacing there. Well, at least it motivates you to keep the actual code lines short.

[kode]

It made sense at the time.

Actually, and sadly, it still makes sense now.

The whitespace is because flashdevelop indents in tabs, and this was already tabbed in a great deal due to class/package/other functions.

The problem isn't the tabs: the problem is having a tab spacing of 8. Cut that to 4 (or 2, if you prefer) and you'll be a much happier chipmunk.

The whole point in using tabs (I personally don't, but ...) is that you, the programmer, can choose where your tab stops are. You want a tab spacing of 4? (setq tab-width 4)

So I say, let him use tabs if he wants. It's not like it's hard to "fix" it