Hi there,

can anyone tell me why programs compiled with -o keep hanging, if I use template set<>:

-SNIP-

set<float> floatcache;
float x, y, z;

vector<string>::iterator vecPos;

if (temp.size() > 0) {

   cout << "scanning for floats... " << endl;

   for (vecPos = temp.begin();
	     vecPos != temp.end();
	     vecPos++) {

	 if (sscanf(vecPos->c_str(), "v %f %f %f", &x, &y, &z) == 3) {

	    	floatcache.insert(x);
		floatcache.insert(y);
		floatcache.insert(z);

	 };

   };

};

-SNIP-

--------------------------------------

if I compile this program without -o,
it runs just fine and inserts all the scanned floats
into the set.

compiler: digitalmars 8.27 / SGI STL 3.3

I downloaded the SGI STL 3.2 from the homepage earlier... in the sample code for set<>, there's a #pragma message saying "JAK! full optimization causes hang!".

This seems to be a know bug. Is there a solution for it?

-elderic

At the moment, there is no resolution for it, although you can try 8.28 out on it. Also, it may not be hanging, it just may be taking a long time.


"Marcus Elderic Koenig" <elderic@t-online.de> wrote in message news:Xns9215B9C9DE6C8marcuselderickoenig@63.105.9.61...
> Hi there,
>
> can anyone tell me why programs compiled with -o keep hanging, if I use template set<>:
>
> -SNIP-
>
> set<float> floatcache;
> float x, y, z;
>
> vector<string>::iterator vecPos;
>
> if (temp.size() > 0) {
>
>    cout << "scanning for floats... " << endl;
>
>    for (vecPos = temp.begin();
>      vecPos != temp.end();
>      vecPos++) {
>
> if (sscanf(vecPos->c_str(), "v %f %f %f", &x, &y, &z) == 3) {
>
>     floatcache.insert(x);
> floatcache.insert(y);
> floatcache.insert(z);
>
> };
>
>    };
>
> };
>
> -SNIP-
>
> --------------------------------------
>
> if I compile this program without -o,
> it runs just fine and inserts all the scanned floats
> into the set.
>
> compiler: digitalmars 8.27 / SGI STL 3.3
>
> I downloaded the SGI STL 3.2 from the homepage earlier... in the sample code for set<>, there's a #pragma message saying "JAK! full optimization causes hang!".
>
> This seems to be a know bug. Is there a solution for it?
>
> -elderic
>