FiveTech Support Forums

Dear Vilian

I've reviewed the latest changes made to classes that may affect this issue, and I haven't seen any significant changes that could alter the behavior compared to previous versions.

Can you provide me with a .sql file that can create the table (including the structure you use) and with some example rows?

Thanks

vilian wrote: Dear Cristobal,

The type of the fields is:

QTE DECIMAL(11,3)
VALOR DECIMAL(12,4)

vilian wrote: Hi Guys,
IF I run the code bellow with my previous version of FWH. I can see the value 3.3732. But if i run this same code with the new version of FWH, the value showed is 3.37. Do you know why ?

Code: Select all Collapse

function Main()
LOCAL oRs,oCn,oError,oQry,cQryMat

   SET DATE BRIT
   SET CENTURY ON
   
   CursorWait()

   oCn := Maria_Connect( aStr[1] )

   oRs := oCn:Query("SELECT nfiscal,cmat,qte,qtedig FROM consumo WHERE nfiscal LIKE '000005829%'")
   cQryMat := ""
   DO WHILE .NOT. oRs:Eof()
      cQryMat += "'"+oRs:cmat+"',"
      ? oRs:qte,oRs:qtedig,oRs:cmat
      oRs:Skip()
   ENDDO
   oRs:GoTop()
   
RETURN NIL

Yes.

I sent it again to your email. please, let me know if you need something else.

We need a little sample to reproduce the problem. The sample must be compilable and runnable without any not available external dependencies (database included).

Hi Enrico,

I sent it by email to Cristobal.

Ok, so you have to wait for Cristobal's answer.

Dear Antonio,

I'm using Borland/32bits.

Antonio Linares wrote: Dear Vilian,

Thanks to Hakan Öztürk as has found the difference in the code :D

It seems as Mr. Rao changed a call to hb_valStrnToNum() into a call to hb_strnToNum()

According to Mr. Rao's notes:

Code: Select all Collapse

#ifdef REMOVED
/*
               hb_arraySetND( aRow, i + 1, ( mrow[ i ] == NULL ? 0.0 : hb_strVal( mrow[ i ], aLen[ i ] ) ) );
               break;  // loss of accuracy in case of bigint large numbers
*/

/*
               // nor working in case of double datatpe with large numbers
               if( ! hb_valStrnToNum( mrow[ i ], aLen[ i ], &lValue, &dValue, &iDec, &iWidth ) )
               {
                  hb_arraySetNL( aRow, i + 1, ( long ) lValue );
               }
               else
               {
                  hb_itemPutNDLen( itemTmp, dValue, iWidth, iDec );
                  hb_arraySetForward( aRow, i + 1, itemTmp );
               }

               break;
*/
#endif

The modified code uses this now:

Code: Select all Collapse

            case 'N':

               if ( ( nAt = ( int ) hb_strAtI( "e", 1, mrow[ i ], aLen[ i ] ) ) > 0 )
               {
                  dValue = hb_strVal( mrow[ i ], aLen[ i ] );
                  x = ( int ) hb_strVal( mrow[ i ] + nAt, aLen[ i ] - nAt );
                  dValue *= pow( ( double ) 10.0, ( double ) x );
                  hb_arraySetND( aRow, i + 1, dValue );
               }
               else
               {
                  if ( hb_strnToNum( mrow[ i ], aLen[ i ], &lValue, &dValue ) )
                  {
                     hb_arraySetND( aRow, i + 1, dValue );
                  }
                  else
                  {
                     hb_arraySetNLL( aRow, i + 1, lValue );
                  }
               }
               break;

We can revert this change and send you a modified lib for your testing. Please let me know if you want it for 32 or 64 bits ? Borland or MSVC ? many thanks!