Hi Viktor
I see that you have vastly improved the fclass1, ffile1, ft_funcs by getting
rid of most of the chr(10), chr(13) mess and using hb_OSNewLine(). I haven't
build the new system yet but see a problem for systems that use anything but
CRLF as the EOL terminator.
looking at the code:
cB := ""
nSavePos := FSEEK( nH, 0, FS_RELATIVE )
nNumRead := FREAD( nH, @cLine, nMaxLine )
IF ( nEol := AT( hb_OSNewLine(), SUBSTR( cLine, 1, nNumRead ) ) ) == 0
cB := cLine
ELSE
cB := SUBSTR( cLine, 1, nEol - 1 )
FSEEK( nH, nSavePos + nEol + 1, FS_SET )
ENDIF
Say the line is as follows:
Pos
1234567890123
xxxxLCxxxxxLC where L is LF and C is CR
If nSavePos = 1, nEol = 5 you seek to 1+5+1=7 which is correct. However for
systems with only LF or CR as EOL you still seek to 7 which is incorrect.
It seems you should seek to:
FSEEK( nH, nSavePos + len( hb_OSNewLine() ) - 1, FS_SET )
1234567890123
xxxxLxxxxxxLx
nSavePos =1, nEOL = 5 len( hb_OSNewLine() ) - 1 = 0
So you now seek to Pos 6 for LF systems and Pos 7 for MS systems which is
correct. It even works for the 3 and 4 byte EOL of some mainframes.
Additionaly, the chr(13) stripping at line 911 isn't needed. It should have
always been chr(10) anyway to remove the dangling LF from the CRLF that was
always left in the way.
FUNCTION FT_FReadLn()
RETURN FReadLn()
_______________________________________________
Harbour mailing list
Harbour@harbour-project.org
http://lists.harbour-project.org/mailman/listinfo/harbour
