Alain Pastor wrote:
> 
OOPS there was a bug in FN REPLACE5$:
the following line:

 P.nil`` = oldLen - ( found + oldSubStr[0] )

should have been:

P.nil`` = oldLen - ( myStr[0] + oldSubStr[0] )

Here is another entry that tries to do the search in line rather
than calling the INSTR function, so far it is the fastest that I
could come up with, if there is no bug in it that is:

LOCAL FN REPLACE8$( myStr AS STR255, oldSubStr AS STR255, newSubStr
AS STR255 )
DIM P       AS POINTER
DIM found   AS LONG
DIM oldLen  AS LONG
DIM i       AS LONG

IF oldSubStr[0] = _nil THEN EXIT FN
found  = 0
P      = @myStr
oldLen = P + P.nil`` - oldSubStr[0]
LONG IF oldLen > 0
"start over"
DO
P++
UNTIL P.nil`` = oldSubStr[1] OR P = oldLen
FOR i = 2 TO oldSubStr[0]
P++
IF P.nil`` != oldSubStr[i] THEN EXIT FOR
NEXT
LONG IF i <= oldSubStr[0] AND P < oldLen
GOTO "start over"
XELSE
found = P - i - @myStr + 2
END IF
LONG IF found
P       = @myStr
oldLen  = P.nil``
P.nil`` = found - 1
P      += P.nil`` + oldSubStr[0]
P.nil`` = oldLen - ( myStr[0] + oldSubStr[0] )
myStr  += newSubStr + PSTR$( P )
END IF
END IF

END FN = myStr

I pity our poor Japanese friends who have to take care of that two
bytes thingy as Osamu pointed out.
Relying upon the FB statement is more secure for that kind of thing.
-- 

Cheers

Alain

-----------------------------------------------------
FB^3 in Europe:  http://euro.futurebasic.com/
FB II Pouch:     http://www.pixmix.com/FB/outils.html
-----------------------------------------------------