Migrating to Harbour

[Antonio Linares]

Enrico,

With my change your code is working fine here

I don't understand why it does not work fine for you there

[Antonio Linares]

Also, I remind you that MemoLine() is very slow. So if you plan to get the lines of a large text, you should avoid it.

We have code in FWH for reading text lines much much faster

[Enrico Maria Giordano]

Antonio,

Enrico,

With my change your code is working fine here

I don't understand why it does not work fine for you there

It's working, but it's very slow as it doesn't use nPos parameter (reading the source code, it's not supported).

EMG

[Enrico Maria Giordano]

Antonio,

Also, I remind you that MemoLine() is very slow. So if you plan to get the lines of a large text, you should avoid it.

We have code in FWH for reading text lines much much faster

It's very fast in xHarbour using nPos parameter. I'm not going to change all my code only to test Harbour, sorry.

EMG

[Rick Lipkin]

Enrico

I have been using some similar ( xHarbour ) code to pull out text stored in a Sql Memo field .. I do not know if it would make any difference in Harbour or not or whether it would be faster. As you can see, I do not pass any other parameters to MemoLine other than the text, the number of characters I wish to print and the line.

Rick Lipkin

Code: Select all

cComment := oRs:Fields("Comment"):Value

// Print comments            
nLinCnt := MlCount( cComment, 70 )    // 110
IF nLinCnt > 0
   FOR x := 1 to nLinCnt
       cText := MemoLine( cComment, 70, x )  // 110
       oPrint:Say( LINE,(oPrint:nHorzRes()*.02), cText, oFont10 )  
       Line += oFont10:nHeight
    NEXT
Endif


 

[Enrico Maria Giordano]

Rick,

I know, but it would be very slow.

EMG

[TimStone]

EMG,

I use Harbour with FWH and MSVC 2013 ... and I have absolutely no speed issues with memos.

My clients add labor ( often a significant amount ) descriptions to invoices, plus all the disclaimers, etc. are all done with memo text ( memo fields ).

The display is instantaneous. The printing is also incredibly fast.

I don't know what you are doing, but as Antonio said, I see no reason why it should be slow.

Tim

[Enrico Maria Giordano]

Tim,

this is a sample:

Code: Select all

FUNCTION MAIN()

    LOCAL cTxt := REPLICATE( "This is a test" + CHR( 13 ) + CHR( 10 ), 10000 )

    LOCAL cNew := ""

    LOCAL nLines

    LOCAL nSec

    LOCAL i

    nLines = MLCOUNT( cTxt, 1024, , .F., .T. )

    ? "Lines:", LTRIM( STR( nLines ) )
    ?

    nSec = SECONDS()

    FOR i = 1 TO nLines
        cNew += ALLTRIM( MEMOLINE( cTxt, 1024, i, , .F., .T. ) ) + CHR( 13 ) + CHR( 10 )
    NEXT

    ? "Seconds:", SECONDS() - nSec

    nLines = MLCOUNT( cNew, 1024, , .F., .T. )

    ? "Lines:", LTRIM( STR( nLines ) )

    RETURN NIL

With parameter nPos (supported by xHarbour) it lasts 0.00 seconds.

EMG

[TimStone]

Enrico,

I will admit I do not test for elapsed time. That is way too theoretical for me. I look at "real world" speeds and that is judged by my clients evaluation of performance.

So, for printing a memo text:

FOR lx := 1 TO MLCount( oTrec:rectxt, 60,, .t. )
PAGEHEAD( oPrn, aHead )
oPrn:Say( nRow, 2 * nCsp, MEMOLINE( oTrec:rectxt, 60, lx,, .t. ), oFnorm )
nRow += nRsp
NEXT

Where rectxt is the memo field in the record. I find that this certainly works faster than a laser printer can push it out.

As for looking at, and editing, the text in a record, I have a multi-line get that I use with the MEMO variable. It is "instant loading". That means, as I browse up and down a list, the memo is displayed ( along with the other data fields ) in the upper half of the dialog. The display in instantaneous.

Tim

[Antonio Linares]

Enrico,

Here is the source code for MemoLine():

https://github.com/harbour/core/blob/ma ... /mlcfunc.c

/* MemoLine( <cString>, [ <nLineLength>=79 ],
* [ <nLineNumber>=1 ],
* [ <nTabSize>=4 ], [ <lWrap>=.T. ],
* [ <cEOL>|<acEOLs> ] ) -> <cLine>
*/

So it seems as it does not support the same parameters as xHarbour. We could compare xHarbour and Harbour and propose to enhance such function as you need it.

[Enrico Maria Giordano]

Tim,

I will admit I do not test for elapsed time. That is way too theoretical for me.

It's not theoretical for me. It's been the solution to a specific slowness problem.

EMG

[Enrico Maria Giordano]

Antonio,

We could compare xHarbour and Harbour and propose to enhance such function as you need it.

Thank you!

EMG

[Antonio Linares]

Enrico,

Now we need to get the xHarbour MemoLine() source code

Do you have it at hand ?

[Enrico Maria Giordano]

Antonio,

here it is:

Code: Select all

HB_FUNC( MEMOLINE )
{
   HB_SIZE        ulLen, ulLineLength, ulTabSize;
   BOOL           fWordWrap;
   PHB_EOL_INFO   pEOLs;
   int            iEOLs;
   const char *   pszString = hb_mlGetParams( 1, &ulLen, &ulLineLength,
                                              &ulTabSize, &fWordWrap,
                                              &pEOLs, &iEOLs );
   char *         szLine;
   ULONG          ulLine   = hb_parnl( 3 );
   HB_SIZE        ulEnd, ulOffset = ISNUM( 7 ) ? hb_parnl( 7 ) - 1 : 0;
   HB_SIZE        ulCols   = 0;

   if( ! pszString )
   {
      hb_retc( NULL );
      return;
   }

   if( ulLine == 0 )
   {
      ulLine = 1;
   }
   while( --ulLine && ulOffset < ulLen )
   {
      ulOffset = hb_mlGetLine( pszString, ulLen, ulOffset,
                               ulLineLength, ulTabSize, 0, fWordWrap,
                               pEOLs, iEOLs, &ulCols, NULL );
   }

   ulEnd = ulOffset;

   if( ulOffset < ulLen )
   {
      ULONG ulCol = 0;
      ulEnd    = hb_mlGetLine( pszString, ulLen, ulOffset,
                               ulLineLength, ulTabSize, 0, fWordWrap,
                               pEOLs, iEOLs, &ulCols, NULL );
      szLine   = ( char * ) hb_xgrab( ulLineLength + 1 );

      while( ulCol < ulCols )
      {
         if( pszString[ ulOffset ] == HB_CHAR_HT )
         {
            HB_SIZE ul = ulTabSize - ( ulCol % ulTabSize );
            do
            {
               szLine[ ulCol++ ] = ' ';
            }
            while( --ul && ulCol < ulCols );
         }
         else if(  pszString[ ulOffset ] == HB_CHAR_SOFT1 &&
                  pszString[ ulOffset + 1 ] == HB_CHAR_SOFT2 )
         {
            ulOffset++;
         }
         else
         {
            szLine[ ulCol++ ] = pszString[ ulOffset ];
         }
         ulOffset++;
      }
      if( ulCols < ulLineLength )
      {
         memset( szLine + ulCols, ' ', ( size_t ) ( ulLineLength - ulCols ) );
      }
      szLine[ ulLineLength ] = 0;

      hb_retclen_buffer( szLine, ulLineLength );
   }
   else
   {
      hb_retc( NULL );
   }
   hb_xfree( pEOLs );

   if( ISBYREF( 7 ) )
   {
      hb_stornl( ( LONG ) ulEnd + 1, 7 );
   }

}

EMG

[Antonio Linares]

Enrico,

have you compared it with Harbour's one ?