FiveTech Software tech support forums

Why can't you use your own function for this?

EMG

Because i not found the way to replace Upper() by my own Upper() in xHarbour, and ADS reconize only "Upper" name in indexed keys.
My function Maj() do what Clipper do, and it is not a problem when i don't use ADS. I cannot use ADS only because this.

I think ADS would use its own function anyway.

EMG

But if ADS is not connected, or if we want change RDD to DBFCDX for example, Upper() of xHarbour is used, who is not compatible with alphabétic order. Because the ASCII codes of all accented chars are after the ASCII code of "Z". This is why it is necessary to "destroy" all accents in Upper() fonction. Upper() of xHarbour gives the real upper chars, but this function would be UpperReal(), not Upper(). Simply Restore Upper() like Clipper do, to keep FULL compatibility in all RDD without changes in the code.

Solved in great part by insert RDDInit() on the start of main .prg,
after reminder by David A. Smith in xHarbour forum.

Upper() continue to return accents in uppercase, but my alphabetic lists are very better. The "é" was after "z" before, now the "é" is before "e" and the "è" is after "e". This is not perfect but acceptable i think for my users.

Best Regards

Great! Can you share the solution, please?

EMG

Mr. Badara Thiam

I am sorry, I am unable to agree.
If we set codepage correctly, sorting and indexing by (x)Harbour is correct and respects the rules of the language set by the codepage.

It is NOT correct that (x)Harbour sorts alphabets by ASCI values. NO. It sorts the alphabets by the language collation rules.

This is the sample to test sorting of French alphabets:

Code: Select all  Expand view

function FrenchCollation()

   local cList, cSort, aList, n

   HB_LangSelect("FR")
   HB_SetCodePage("FRWIN")

   cList := "azertyuiopéqsdfghjklmùwxcvbnèà"
   ? cList, Upper( cList )

   aList    := {}
   for n := 1 to Len( cList )
      AAdd( aList, { SubStr( cList, n, 1 ), ,  } )
      aList[ n, 2 ]  := ASC( aList[ n, 1 ] )
      aList[ n, 3 ]  := UPPER( aList[ n, 1 ] )
   next

   ASort( aList, , , { |x,y| x[ 1 ] < y[ 1 ] } )
   XBROWSER aList TITLE "FRENCH SORTING" ;
      SETUP ( oBrw:cHeaders := { "CHAR", "ASC", "UPPER" }, ;
              oBrw:nDataStrAligns := AL_CENTER ) ;
      SHOW RECID

return nil
 

Let us now test indexing. Please remember to USE dbf with CODEPAGE clause.

Code: Select all  Expand view

function frenchIndexing()

   field FLDCHR, FLDASC
   local cList, n

   HB_LangSelect("FR")
   HB_SetCodePage("FRWIN")

   cList := "azertyuiopéqsdfghjklmùwxcvbnèà"
   ? cList, Upper( cList )

   FERASE( "TFRENCH.CDX" )
   DBCREATE( "TFRENCH.DBF", { { "FLDCHR", "C", 1, 0 }, { "FLDASC", "N", 5, 0 } }, "DBFCDX" )
   USE TFRENCH NEW EXCLUSIVE VIA "DBFCDX" CODEPAGE "FRWIN"
   for n := 1 to Len( cList )
      DBAPPEND()
      FLDCHR := SubStr( cList, n, 1 )
      FLDASC := ASC( FLDCHR )
   next
   INDEX ON UPPER(FLDCHR) TAG FLDCHR
   GO TOP
   XBROWSER "TFRENCH" TITLE "FRENCH DBFCDX COLLATION" ;
      COLUMNS "FLDCHR", "FLDASC", "ORDKEYVAL()" SHOW RECID ;
      SETUP ( oBrw:nDataStrAligns := AL_CENTER )

return nil

Both Harbour and xHarbour produce the same results.
I do not think there is anything wrong with the above collation.
I think we can safely rely on (x)Harbour.

Dear G. N. Rao,

I could not use your browser because there is a « syntax error at SHOW » for my xHarbour compiler.
Here are lists with my internal browser, printed in pdf :

- http://icim.fr.free.fr/test/test-french ... -fr850.pdf
- http://icim.fr.free.fr/test/test-french ... -frwin.pdf

As you can see, it is better with FR850.
I have seen that the CODEPAGE parameter of USE is ignored. Only HB_SetCodePage() change something.

I would point out that all the chars in my databases are not based on ANSI but on OEM, MS-DOS CP 850.
Since 1986 with Clipper. All that is put on the screen or printer with my Windows applications,
is translated before with OemToAnsi().

I would just create a separate field in your database where the accents are eliminated and use that for the index key.

It is more than 30 years since 1986. Lot has changed since then. For all those who are maintaining data in ANSI, as is the present (default) practice, (x)Harbour default behaviour works perfectly.

One option available is to migrate the data to ANSI and use contemporary tools comfortably.

In ADS also, we can maintain data in ANSI and choose our preferred language collation.